The association between levosimendan and mortality in patients with sepsis or septic shock: a systematic review and meta-analysis

Abstract

Background

Levosimendan is increasingly being used in patients with sepsis or septic shock because of its potential to improve organ function and reduce mortality. We aimed to determine if levosimendan can reduce mortality in patients with sepsis or septic shock via meta-analysis.

Evidence sources and study selection

We comprehensively searched the PubMed, Embase, Web of Science, and Cochrane Library databases from inception through 1 October 2022. Literature evaluating the efficacy of levosimendan in patients with sepsis or septic shock was included.

Data extraction and outcome measurements

Two reviewers extracted data and assessed study quality. A meta-analysis was performed to calculate an odds ratio (OR), 95% confidence intervals (CI), and P-values for 28-day mortality (primary outcome). Secondary outcomes included changes in indexes reflecting cardiac function before and after treatment, changes in serum lactate levels in the first 24 h of treatment, and the mean SOFA score during the study period. Safety outcomes included rates of tachyarrhythmias and total adverse reactions encountered with levosimendan.

Results

Eleven randomized controlled trials were identified, encompassing a total of 1044 patients. After using levosimendan, there was no statistical difference between groups for 28-day mortality (34.9% and 36.2%; OR: 0.93; 95% CI [0.72–1.2]; P = 0.57; I² = 0%; trial sequential analysis-adjusted CI [0.6–1.42]) and sequential organ failure assessment (SOFA) score, and more adverse reactions seemed to occur in the levosimendan group, although the septic shock patient’s heart function and serum lactate level improved.

Conclusion

There was no association between the use of levosimendan and 28-day mortality and SOFA scores in patients with septic shock, though there was statistically significant improvement in cardiac function and serum lactate.

Introduction

Sepsis and septic shock are some of the major causes of death worldwide, there were 48.9 million new sepsis patients worldwide and 11 million deaths due to sepsis, accounting for 19.7% of global deaths in 2017 [1]. About 50% of sepsis patients will have varying degrees of myocardial injury, called septic cardiomyopathy (SCM) causing deterioration of hypoperfusion. Its pathogenesis may be related to hemodynamics and myocardial changes, mitochondrial fission, autophagy and apoptosis of myocardial cells, calcium imbalance, and so on [2–4]. Septic shock patients with SCM are more likely to face irreversible shock, thus increasing the risk of death by up to 70% [5]. At present, the main clinical treatment for SCM is to use positive inotropic drugs while concurrently fighting the underlying infection compared with other positive inotropic drugs or use it to improve microcirculation.

Levosimendan is a calcium sensitizer that can reduce diastolic relaxation and regulate blood circulation and cardiovascular function without significantly increasing the demand for oxygen supply [6]. It plays a beneficial role in supporting and protecting the heart under different pathological conditions, which has been verified in mice and humans [7–9]. Its mechanism has not been completely clarified in septic shock, including the inhibition of inducible nitric oxide synthase gene expression [9]. Studies have shown that levosimendan can improve liver and kidney functions, as well as improve systemic hemodynamics and increase splanchnic perfusion [10,11]. Compared with dobutamine, which has been recommended in the past, levosimendan has some advantages in sepsis and septic shock, such as not increasing myocardial oxygen consumption or the risk of malignant arrhythmias. Some studies have shown that levosimendan may reduce mortality in patients with septic shock [11,12]. Improvement in heart function and other organ functions has also been demonstrated [13–15]. However, a randomized controlled trial (RCT) of 515 patients suggested that levosimendan could not reverse organ dysfunction or mortality, increased the time needed for mechanical ventilation, and increased the incidence of arrhythmias [16]. Since then, there has been a plethora of studies seeking to determine the specific role of levosimendan in sepsis or septic shock. Up to now, the role of levosimendan in sepsis or septic shock is still questionable, and whether it has more advantages than traditional positive inotropic drugs is not completely determined so we conducted this meta-analysis.

Methods

This systematic review and meta-analysis were conducted according to the Cochrane Handbook for Systematic Reviews of Interventions [17] and written based on the Preferred Reporting Items for Systemic Reviews and Meta-analysis guidelines [18]. A prospective protocol was prepared, including objectives, strategies to search the literature, inclusion and exclusion criteria, methods for statistical analysis, and lists and measurements of outcomes.

Strategies to search the literature

A literature search was performed in the electronic databases of PubMed, Embase, Web of Science, and Cochrane Library from their inception dates to 1 October 2022, unrestricted to region, language, publication type, or year, using the keywords levosimendan, sepsis, and septic shock to identify published RCTs and observational studies evaluating the efficacy of levosimendan for sepsis or septic shock patients.

Inclusion criteria

Literature was included based on the following criteria: patients: age ≥18 years. Sepsis definition: suspected or confirmed infection with sequential organ failure assessment (SOFA) score of two or higher. Septic shock definition: sepsis with persistent hypotension requiring vasopressors to maintain a mean arterial pressure ≥65 mmHg and a serum lactate level >2 mmol/L despite adequate fluid challenge [19]. If the target literature did not mention these definitions or use sepsis definitions not belonging to Sepsis-3, our team would meet to evaluate whether it should be included.

Only RCTs comparing levosimendan to placebo or other treatment groups that provided at least one outcome would be considered. Exclusion criteria were randomized trials or retrospective comparative studies without a placebo or treatment group, or studies with insufficient data.

Data extraction and available outcomes of interest

All available data including first author, publication year, country of origin, number of participants, participant characteristics, interventions, and outcomes were extracted, entered and summarized independently by screening title and abstract or full text by two investigators. Any disagreement was resolved by consensus. 28-day mortality was the primary outcome. Secondary outcomes included changes of indexes reflecting cardiac function and myocardial injury before and after treatment including cardiac index (CI), left ventricular stroke work index (LVSWI), left ventricular ejection fractions (LVEF), N-terminal pro-B-type natriuretic peptide (NTproBNP) [20], and changes in serum lactate levels at 24 h and mean SOFA scores during the study period. Based on the mechanism of action of levosimendan, it may produce adverse reactions such as hypotension and tachycardia, so we also collected the total incidence of adverse reactions and the incidence of arrhythmias.

Quality assessment

The methodological quality assessment of the included studies was made by two investigators. The methodological quality of RCTs was based on the Cochrane risk of bias tool which consists of seven items: sequence generation, allocation concealment, blinding of participants/personnel, blinding of outcome assessment, incomplete data outcomes, selective reporting, and other biases [17]. Each quality item was then graded as having low, high, or unclear risk. Any disagreement was resolved by an adjudicating author.

Statistical analysis

All meta-analyses were performed using Review Manager 5.2 (Cochrane Collaboration, Oxford, UK) and trial sequential analysis v.0.9.5.10 Beta [21]. The weighted mean difference (WMD) and odds ratio (OR) were used to compare continuous and dichotomous variables, respectively. When we encounter a value of 0, we will use risk difference (RD) instead of OR in Review Manager 5.2 for calculation. Continuous data were represented as medians with interquartile ranges, and means with standard deviations were calculated by methods reported by Wan et al. [22]. All outcomes were reported with 95% confidence intervals (95% CI), P-values (P), and I². Statistical heterogeneity among different studies was determined by the I² statistic. The random-effects model was used to pool the data if there was heterogeneity between studies (defined by I² > 50% or P-value < 0.1 of Q-test in Review Manager); otherwise, the fixed-effects model was used [17]. Publication bias was assessed by funnel plots if the number of studies reporting the primary outcomes was 10 or more. All tests were two-tailed, and P < 0.05 was considered statistically significant.

We performed a trial sequential analysis to assess the increased risk of random errors for mortality if there was relatively sparse data and repeated significance testing. The result was displayed on a trial sequential analysis diagram with a trial sequential analysis-adjusted CI and an adjusted level of statistical significance. Trial sequential analysis is used to reduce the risk of a wrong conclusion in a meta-analysis that did not achieve the required information size (RIS). Trial sequential analysis-adjusted CI was calculated by the random-effect model for diversity (D²) with a 5% risk of type I error and a power of 80%. For the estimate of the RIS, we set the intervention effect of a 15% relative risk reduction and calculated the control event incidence from the conventional meta-analysis.

Results

A total of 565 articles were screened (136 from Pubmed, 233 from Embase, 161 from Web of Science, and 35 from Cochrane Library). Of these identified studies, titles and abstracts were screened for inclusion. Finally, the full text of 24 records was reviewed, and 11 met inclusion criteria (Supplementary File 1, Supplemental digital content 1, http://links.lww.com/EJEM/A412). The research directed by Antcliffe et al. [23] was a substudy of Leopards so the non-repetitive outcomes of the former were integrated into Leopards by Gordon et al. [16]. Unfortunately, two studies from China were rejected at the final stage because they could not obtain complete data [6,24]. There were 522 cases in the experimental group (levosimendan) and 522 cases in the control group (dobutamine or placebo), a total of 1044 cases.

Characteristics of included literature

Research on levosimendan and septic shock has been carried out all over the world (Table 1). Eight studies compared the efficacy of levosimendan or dobutamine in the treatment of sepsis or septic shock, and three of 11 were levosimendan compared to placebo [12,16,25]. All the studies were conducted on the topic of septic shock. Almost all studies found that the main source of infection was pulmonary, followed by intra-abdominal infection. Three studies were aimed at low cardiac output (LVEF ≤ 45%/50%) which may be related to SCM [14,26,27]. Due to various reasons, the remaining studies did not grade the cardiac function of the enrolled patients and considered the potential anti-inflammatory and microcirculation improvement effects of levosimendan. Studies such as Gordon et al. conducted a subgroup analysis of people with low cardiac output [16]. Levosimendan dosing in almost all works of literature was 0.1–0.2 μg/kg/min while dobutamine was 5 μg/kg/min.

Table 1.

Characteristics of included trials

Study Year	Country	Sample size (E/C)	Male Gender (E/C)	Mean/Median* age (E/C)	Intervention(E)	Control(C)	Main infectious site	Inclusion criteria	Cardiovascular criteria	Dosage regimen design(E)	Dosage regimen design (C)
Morelli 2006	Italy	13/15	10/11	62.4/61.5	Levosimendan	Dobutamine	NA	SS	LVEF > 45%	0.2 μg/kg/min	5 μg/kg/min
Xu 2018	China	15/15	8/8	87.9/88.1	Levosimendan	Dobutamine	Lung	SS	LVEF ≤ 50%	0.2 μg/kg/min	5 μg/kg/min
Wang 2017	China	120/120	69/65	70/69	Levosimendan	routine therapy	Lung	SS	NA	0.1–0.2 μg/kg/min	NA
Hajjej 2017	Tunis	10/10	9/8	61/51*	Levosimendan	Dobutamine	Lung	SS	NA	0.2 μg/kg/min	5 μg/kg/min
Meng 2016	China	19/19	13/11	55.4/50.2	Levosimendan	Dobutamine	Lung	SS	LEVF ≤ 45%	0.2 μg/kg/min	5 μg/kg/min
Torraco 2014	Italy	13/13	11/8	70/68	Levosimendan	routine therapy	Lung	SS	NA	0.2 μg/kg/min	NA
Fang 2014	China	18/18	14/13	61.4/61.7	Levosimendan+Dobutamine	Dobutamine	Lung	SS	LEVF ≤ 45%	NA	NA
Memiş 2012	Turkey	15/15	8/8	56.3/54.9	Levosimendan	Dobutamine	NA	SS	NA	0.1 μg/kg/min	10 μg/kg/min
Morelli 2010	Italy	20/20	14/13	68/66	Levosimendan	Dobutamine	Lung	SS	NA	0.2 μg/kg/min	5 μg/kg/min
Alhashemi 2009	Saudi Arabia	21/21	NA	NA	Levosimendan	Dobutamine	NA	SS	NA	0.05–0.2 μg/kg/min	5 μg/kg/min
Gordon 2016	UK	258/257	145/144	67/69*	Levosimendan	Placebo	Lung	SS	NA	0.1–0.2 μg/kg/min	NA

Age*, mean ± SD/median and IQR; C, control group; E, experimental group; LVEF, left ventricular ejection fractions; NA, Not available in the included literature; RCT, Randomized clinical trial; SS, Septic shock.

Methodological quality assessment of the included studies

According to the Cochrane risk of bias tool, 10 RCTs showed generally high quality and 1 low quality [28] (Fig. 1). More than 90% of included studies provided random grouping and allocation methods, such as random number tables, computer-generated random allocation, envelope use, etc. Ten-elevenths works of literature were blinded to participants. All the works have completely reported the results of their expected analyses, and there was no selective reporting noted.

Fig. 1.

Risk of bias graph: review authors’ judgments about each risk of bias item presented as percentages across all included RCTs.

Primary outcome

In the 11 included studies, all reported 28-day mortality rates on a total of 1044 adult patients. Mortality was 34.5% in the levosimendan group and 36.2% in the control group in these included studies (OR: 0.93; 95% CI [0.72–1.20]; P = 0.57; I² = 0%; fixed-effect model) (Fig. 2).

Fig. 2.

Forest plot of the primary outcome (mortality).

The trial sequential analysis results demonstrated that the information size needed to detect an intervention effect was 2372 patients (trial sequential analysis-adjusted CI [0.60 to 1.42], moderate-certainty evidence). The cumulative Z curve did not cross either the conventional boundary for benefit nor the O’Brien Fleming boundaries for benefit, but crossed the futility area which means the use of levosimendan in the experimental group was not superior to that in the control group and more patients and trials would be necessary in the future. (Fig. 3). That is to say, the current 1044 sample size is still not saturated. In the future, Levosimendan can continue to do relevant research in this area to explore its effect on mortality.

Fig. 3.

The result of trial sequential analysis.

Subgroup analysis

The control groups were divided into placebo or dobutamine. As shown in Supplementary File 2, Supplemental digital content 1, http://links.lww.com/EJEM/A412, levosimendan still cannot decrease mortality when compared to placebo or dobutamine (levosimendan versus placebo: 32.7% and 33.2%, OR: 0.98, 95% CI [0.73–1.32], P = 0.90, I² = 62%; levosimendan versus dobutamine: 39.7% and 45.1%, OR: 0.78, 95% CI [0.47–1.32], P = 0.36, I² = 0%).

Secondary outcomes and safety outcomes

Seven additional outcomes were selected to examine any additional impact of levosimendan on sepsis or septic shock, in which changes of CI, LVSWI, LVEF, and NTproBNP reflected the changes of cardiac function, serum lactate reflected the metabolic stress, and SOFA scores reflected the level of overall organ dysfunction. Data for secondary outcomes are shown in Table 2 and Supplementary File 4, Supplemental digital content 1, http://links.lww.com/EJEM/A412.

Table 2.

Secondary outcomes and safety outcomes of meta-analysis comparison of experimental group and control group

Outcomes of interest	Number in Studies	Experimental cases	Control cases	WMD/RD*	95% CI	P-value	Study heterogeneity
							Chi2	df	I2, %	P-value
Cardiac index	5	80	82	0.42	0.25 to 0.59	<0.00001	7.38	4	46	0.12
LVSWI	4	70	72	4.89	3.70 to 6.07	<0.00001	0.46	3	0	0.93
LVEF	4	65	67	6.41	3.16 to 9.65	0.0001	0.60	3	0	0.90
NTproBNP	2	217	232	−0.12	−0.24 to −0.00	0.04	0.39	1	0	0.53
Serum lactate	6	95	97	−1.00	−1.45 to −0.56	<0.0001	3.70	5	0	0.59
SOFA scores	2	378	377	0.21	−1.44 to 1.87	0.80	2.38	1	58	0.12
Safety outcome
Total adverse reactions	3	291	292	0.03*	−0.02 to 0.08	0.21	0.50	2	0	0.78
Arrhythmia	3	291	292	0.03*	−0.00 to 0.06	0.06	0.53	2	0	0.77

95% CI, 95% confidence intervals; LVEF, left ventricular ejection fractions; LVSWI, left ventricular stroke work index; NTproBNP, N-terminal pro-B-type natriuretic peptide; RD, risk difference; SOFA, sequential organ failure assessment; WMD, weighted mean difference.

Results showed that levosimendan use was associated with the improvement in CI, LVSWI, and LVEF, while NTproBNP and serum lactate decreased more significantly. However, the SOFA score did not improve significantly. Five studies reported the change of CI for 162 patients (WMD: 0.42; 95% CI [0.25 to 0.59]; P < 0.01; I² = 46%). Change of LVSWI was pooled from four studies involving 142 patients (WMD: 4.89; 95% CI [3.7–6.07]; P < 0.01; I² = 0%). Four studies reported the change in LVEF for 132 patients (WMD: 6.41; 95% CI [3.16–9.65]; P = 0.0001; I² = 0%). NTproBNP was reported in two studies with a total of 449 patients by the fixed-effect model (WMD: −0.12; 95% CI [−0.24 to −0.00]; P = 0.04; I² = 0%). Change in serum lactate was reported in six studies with 192 patients (WMD: −1.00; 95% CI [−1.45 to −0.56]; P < 0.01; I² = 0%) by the fixed-effect model and the change in SOFA scores was calculated from two studies with 755 patients (WMD: 0.21; 95% CI [−1.44 to 1.87]; P = 0.80; I² = 58%) by the random-effect model.

The total incidence of adverse events and the incidence of arrhythmia were used to evaluate the safety of levosimendan. The results are shown in Table 2. It can be seen that the use of levosimendan seemed to increase the occurrence of adverse reactions (11.3% and 8.2%; RD: 0.03; 95% CI [−0.02 to 0.08]; P = 0.21; I² = 0%) and increases the risk of arrhythmia (5.5% and 2.4%; RD: 0.03; 95% CI [−0.00 to 0.06]; P = 0.06; I² = 0%).

Sensitivity analysis and publication bias

We excluded potentially low-quality literature from our sensitivity analysis. Results showed that the significance of any outcome did not change with this exclusion (see Supplementary File 3, Supplemental digital content 1, http://links.lww.com/EJEM/A412). As shown in Supplementary File 5, Supplemental digital content 1, http://links.lww.com/EJEM/A412, no publication bias was found in the included studies.

Discussion

The results of this meta-analysis of 1044 patients from 11 RCTs showed that the use of levosimendan did not decrease 28-day mortality and improve the overall organ function, though improved cardiac function and reduced the production of lactate.

As an inotropic drug, levosimendan has been used in cardiovascular system research for nearly 20 years, but its mechanism of efficacy in sepsis or septic shock has not been fully elucidated yet. It increase in myocardial contractility is achieved with a minimal increase in myocardial oxygen demand and the stabilization of Ca2+ concentration in myocardial cells which will not impair cardiac diastolic function [29–31]. With the long half-life of its active metabolite, a single 24-hour infusion can provide hemodynamic effects for 1 week, which may be sufficient to support most patients with septic shock until hemodynamic recovery [16,29,32]. In addition to the inotropic effect mentioned above, levosimendan also has anti-inflammatory [33], antioxidant [34], anti-apoptosis, and other effects [35], which is why levosimendan appears so promising in the field of sepsis research. This is also why many researchers did not set up cardiac dysfunction when screening for inclusion in the population.

In addition, in clinical practice, the use of levosimendan in patients with septic shock does not necessarily require evidence of low cardiac output. Because more than 30% of patients with septic shock will have some degree of cardiac dysfunction, levosimendan was theorized to be potentially helpful in all sepsis patients, even if they did not have cardiac output deficiencies [16,36]. Out of the included studies, 8 studies did not require evidence of low cardiac output before using levosimendan. Many investigations, including this study, have proved that levosimendan can improve cardiac function in shock patients, mainly manifested in the improvement of ejection fraction and cardiac output [14,23,27].

This meta-analysis also showed that the use of levosimendan can not reduce SOFA scores. Some previous studies found that levosimendan not only improves heart function, but also has potential benefits for liver, kidney, and lung functions, such as decreased bilirubin level, less need for renal replacement therapy, and improved oxygenation indices [11,15,37]. Later, Gordon et al. believed that levosimendan had no beneficial effect on the overall SOFA score or any single part of the score [16]. This is also a difficult problem for all scientists. After using levosimendan, cardiac function has improved and lactate levels have decreased, indicating that microcirculation perfusion may have been repaired, but overall organ dysfunction has not improved significantly. The reasons are comprehensive and complex. For example, the inflammatory factor storm in patients with sepsis is systemic, and various organs may be damaged. Although levosimendan has potential anti-inflammatory and anti-apoptotic effects, its efficacy or its leading role may not be well exerted [14,15].

Although levosimendan has many advantages, it still cannot improve the most important outcome of sepsis and septic shock patients - mortality. This may be affected by the included sample size, as our trial sequential analysis confirmed that 1044 sample sizes did not reach saturation. Some researchers believe that this may be related to a lack of screening for cardiac dysfunction in included patients, but subsequent research and other studies on low cardiac output have denied this view [14,16,26,38]. Of course, identifying SCM is necessary in clinical practice. If we conduct a larger sample size study on SCM, the results may change. The results from our subgroup analysis showed that levosimendan was not superior to dobutamine or placebo in reducing mortality. In general, it is more likely that septic shock, being a serious stress response of the host to infection, cannot effectively improve mortality simply by using a single drug, much like recent controversies involving thiamine or vitamin C therapy for septic shock in recent years [39,40]. If we want to fundamentally improve mortality rates in sepsis, we probably need to make more than one change.

We also conducted a meta-analysis of the adverse reactions of levosimendan for the treatment of sepsis. Levosimendan may lead to a decrease in patient’s blood pressure by expanding pulmonary arteries and other organs’ arteriovenous vessels [41]. So that in patients with shock, they seemed to use higher doses of vasoactive drugs such as norepinephrine [13,16]. The heart rate in the levosimendan group seemed to be higher than that in the placebo or dobutamine groups, and the same occurred in the incidence of tachyarrhythmias in the levosimendan group. These adverse reactions are probably caused by a series of effects caused by levosimendan, such as vasodilation and catecholamine toxicity [42,43]. However, due to the small sample size, our results may have weak persuasiveness. If more research in the future can focus on the incidence of adverse reactions, perhaps this result will be changed.

Limitations

This meta-analysis has a few limitations that should be taken into account. Two included studies provided a larger sample size of 240 patients [25] and 515 patients [16], respectively. Each sample size of the remaining 9 studies was less than 50. Of the eleven included studies, one RCT was with both unknown random sequence generation and allocation concealment. These uncertainties increase the risk of bias in the included studies. Although all the included studies used mortality as a studied outcome for patients, only a small number of studies reported other indicators reflecting the prognosis of sepsis and septic shock patients, such as changes in SOFA score, serum lactate level, or arrhythmias. These facts reflect the heterogeneity of the current levosimendan literature and we have tried our best to reduce heterogeneity and bias in methodology. Some other indicators, such as length of stay in the hospital, duration of ventilator use, and rates of using renal replacement therapy, were not included in this meta-analysis due to a paucity of reporting. Nevertheless, this does not mean that these indicators are not important. Future research should pay attention to adverse reactions as well as mortality and organ dysfunction.

Conclusion

In this systematic review and meta-analysis, the use of levosimendan in patients with septic shock was not associated with the decrease in 28-day mortality and SOFA score, although cardiac function and serum lactate level improved.

Acknowledgements

This work was supported by the National High Level Hospital Clinical Research Funding (2022-PUMCH-B-109) and CAMS Innovation Fund for Medical Sciences (CIFMS) (2021-I2M-1-020).

Study concept and design: Yi Li, Yanxia Gao. Acquisition and analysis of studies and data: Zengzheng Ge and Mubing Qin. Paper writing: Zengzheng Ge. Study revision: All.

All data associated with this manuscript are included in the main text and supplementary materials. If additional information is desired, please contact the corresponding author.

Ethics approval: As this was a meta-analysis, no ethics approval was required.

Conflicts of interest

There are no conflicts of interest.