All Spanish centers with an adult HT program (N=16) were invited to participate in the LEVO-T registry. Inclusion criteria encompassed all patients older than 18 years who were included on the elective waiting list between January 1, 2015 and September 1, 2020. Patients were retrospectively reviewed and followed up until June 2021. Those that received more than 1 dose of ambulatory levosimendan while on the waiting list and before HT were included in the treatment group, while patients not requiring levosimendan were included in the control group. We excluded patients with left ventricular assist devices at the time of inclusion on the waiting list. The decision to administer levosimendan treatment was made by each individual center based on their local criteria, which were established according to prior clinical experience. The Spanish Transplant Organization does not prioritize adult patients on inotropes, who therefore remain on the elective HT list.

The study was approved by the Hospital Universitari de Bellvitge ethics committee, and conformed to the principles of the Helsinki Declaration and with the International Society for Heart and Lung Transplantation (ISHLT) ethics statement. All patients signed an informed consent form for prospective collection of their anonymized data as part of their inclusion in the Spanish Heart Transplant Registry. The authors from each participating center guarantee the integrity of the data. This study was designed in accordance with the Strengthening the Reporting of Observational studies in Epidemiology (STROBE) guidelines.

Data were retrieved by each center from the Spanish Heart Transplant Registry and electronic medical records (see appendix 2). The baseline evaluation was conducted at the time of inclusion on the HT waiting list. Variables included demographic characteristics, HF symptoms, comorbidities, previous HF admissions, medical treatment, and devices. Complementary tests included laboratory tests, echocardiography, and right heart catheterization before inclusion. Among patients receiving levosimendan, additional data were gathered at the time of the first administration and included dose, infusion speed, number of doses, vital signs, laboratory tests, echocardiographic parameters, and right heart catheterization data. Events collected during follow-up included the following: ventricular arrhythmias requiring admission to the emergency department or implantable cardioverter defibrillator therapies for ventricular arrhythmias; HF admissions defined as hospitalization for HF for> 24hours; need for mechanical circulatory support; inclusion on the urgent or emergent waiting list, HT, or death. Among patients who received a HT, additional information including laboratory tests and echocardiographic parameters before HT were collected, as well as data on primary graft dysfunction according to the ISHLT consensus definition,18 need for extracorporeal membrane oxygenation (ECMO), and survival after HT. To ensure the quality of the data, several criteria were established: a) data could only be included if they were within a range of reasonable values for each quantitative variable, previously defined by the research team; b) the dates of the different periods analyzed had to be correlative; c) before the statistical analysis, a search was made for missing and anomalous data and researchers were asked to make a review.

Normally distributed continuous variables are expressed as mean±standard deviation (SD) while nonnormally distributed variables are expressed as median [interquartile range]. Continuous variables are presented as number and percentage for categorical variables. The Kolmogorov-Smirnov test was used to assess normal distribution in continuous variables. The Student t-test and Mann-Whitney U test were used to compare variables with and without normal distribution, respectively. The chi-square test was used to compare categorical variables. The Student t-test for paired samples was used to compare changes in patients’ characteristics during follow-up.

The Kaplan-Meier method was used to describe the survival of patients on the waiting list and after HT. Follow-up was censored at the time of HT or exclusion from the waiting list due to ineligibility. Differences between groups were analyzed with Cox regression models. Since differences were found between the levosimendan group and the control group, we carried out a pairing with the propensity score matching technique to minimize these differences. We adjusted the following covariates (table 1 of the supplementary data): age, sex, ischemic etiology, and the following parameters at the time of inclusion on the waiting list: weight, body mass index, New York Heart Association (NYHA) IV functional class, number of previous HF admissions, implantable cardioverter defibrillator carrier, CRT carrier, use of neurohormonal treatment and diuretics, creatinine values, left ventricular ejection fraction, mean pulmonary arterial pressure, pulmonary capillary wedge pressure, cardiac output, and pulmonary vascular resistance. To perform the pairing, a nonparsimonious logistic regression analysis was first performed, with the dependent variable being “receiving levosimendan” and the independent variables being those previously mentioned. Then, 1:1 pairing was performed and the condition to generate each pair was that the difference in the probabilities was less than 0.024 (20% of the SD of the probabilities, which was 0.12). We then checked that the covariates were balanced between the 2 groups in 2 ways: checking the absence of significant differences (P> .05) and calculating the standardized difference between the 2 groups for each covariate. Small values (≤ 10%) were considered to support the equilibrium assumption in both groups.

The rate of HF admissions was compared between patients who started levosimendan in the first 30 days after inclusion with those who did not with the Student t-test for paired samples. Additionally, a subanalysis of survival and adverse effects was performed between patients with flexible and other dose patterns.

The statistical analysis was performed using SPSS software version 21 (IBM Corp, United States).

Table 1 displays the main baseline characteristics of the cohort at the time of HT listing. The mean age was 54.1±11.2 years and 27.7% were female. The underlying cause of HF was ischemic cardiomyopathy in 308 (30.3%) of patients, 522 (55.3%) had concomitant atrial fibrillation, and most patients (n=580, 57.3%) were in NYHA class IV. Overall, the patients were on optimal medical treatment for HF except for low prescription of sodium-glucose cotransporter 2 inhibitors during the study period (2015-2020). Mean NT-proBNP levels were 5257±16 242 pg/mL, and mean left ventricular ejection fraction was 27.8±13.1%.

Compared with the control group, patients requiring levosimendan had more frequent HF admissions during the previous year, a higher proportion had an implantable cardioverter defibrillator (86.5% vs 74.2%; P <.001), cardiac resynchronization therapy (31.6% vs 24.2%; P=.023), percutaneous edge-to-edge mitral valve repair (6.4% vs 2.4%; P=.003) and higher median (30.4 vs 28.8mmHg; P=.037) and diastolic (22 vs 20.7mmHg; P=.030) pulmonary pressures. Patients in the control group were more frequently female (29.3% vs 22.3%; P=.033), had lower requirements for loop diuretics (92.6% vs 97.0%; P=.016), had less beta-blockers (77.4% vs 84.9%; P=.014), lower levels of serum creatinine (1.25±0.5 vs 1.39±0.6mg/dL; P=.001) and higher left ventricular ejection fraction (28.7±14.0 vs 25.3±9.9%; P <.001). No statistically significant differences were found in age, functional class, or NT-proBNP levels.

Table 2 shows the main features of levosimendan infusion patterns across centers. More than half of the patients started levosimendan within the first 30 days after inclusion (median, 29 days; interquartile range [IQR]: 10-119) and 55.5% had received levosimendan previously in the context of acute HF hospitalization. There was wide heterogeneity in dose patterns, but fixed doses were more frequently prescribed than weight-adjusted doses. The most frequent dose was 6.25mg per session (n=54, 22.7%) followed by 12.5mg per session (n=51, 21.4%) and 0.1μg/kg/min in 24hours (n=50, 21.0%). Most patients received flexible infusions according to clinical requirements during follow-up (n=136, 57.1%) instead of stable regimes (fixed or unlimited). Bi-weekly infusions were used in 49.5% of patients in stable regimes.

Levosimendan infusions were well tolerated with a low proportion of patients (n=5, 2.1%) experiencing adverse events that led to discontinuation. In detail, 2 patients (0.8%) had symptomatic hypotension, 2 patients (0.8%) had nonfatal ventricular arrhythmias, and 1 patient (0.4%) had supraventricular arrhythmias. In addition, 27 patients (11.3%) discontinued levosimendan due to a lack of clinical response, and 12 patients (5%) due to significant clinical improvement. Overall, 20.2% of patients discontinued treatment during follow-up. Discontinuation was less frequent in patients receiving the drug in a flexible-dose pattern compared with other patterns [9 (7.1%) vs 40 (40.2%)], mostly due to futility (table 4 of the supplementary data).

Overall, the patients were on the waiting HT list for a median of 4.2 [1.4-9.1] months. A total of 935 patients (92.1%) received HT during follow-up, 8 patients (0.8%) were on the waiting list at the last evaluation, 33 patients (3.3%) died while on the waiting list, and 39 patients (3.8%) were excluded due to new-onset contraindications (n=22) or due to significant clinical improvement (n=17). No patients were lost to follow-up. Patients who received levosimendan had a significantly longer waiting time for HT (169 [IQR 72-292] days vs 103 [IQR 39-241] days, P <.001).

Figure 2 displays a flowchart of HF admissions according to the time of levosimendan initiation. The proportion of patients requiring hospitalization did not significantly differ between patients who started levosimendan in the first 30 days after inclusion and those who did not (n=39; 33.6% vs n=310; 34.5%; P=.848). In the latter group, almost one-third (n=102; 32.9%) crossed over to the levosimendan group after an HF admission, and the HF admission rate decreased from 0.57 to 0.21 HF admissions per month after levosimendan initiation. Total rates of HF admissions while patients were on the waiting list were similar among those who never received levosimendan, those who initiated the treatment in the first 30 days after inclusion, and those who started the treatment later (0.21, 0.24, and 0.26, respectively).

Figure 2.

Flowchart of heart failure admissions according to time of levosimendan initiation. HF, heart failure; HT heart transplantation; Lev, levosimendan.

(0.25MB).

Overall, 169 patients (16.6%) required mechanical circulatory support while on the waiting list and 133 patients (13.1%) required urgent mechanical circulatory support due to acute cardiogenic shock (35 left Centrimag, 13 biventricular Centrimag, 25 veno-arterial ECMO, 30 intra-aortic balloon pump, and 32 Impella), 36 patients (3.5%) required long-term left ventricular assist devices (LVAD) due to chronic worsening HF. Urgent HT was required in 170 patients and no differences were found between groups in the proportion of patients requiring urgent HT (20.0% vs 17.7%; P=.445). Figure 3 shows 1-year survival on the waiting list among patients who received levosimendan and the control group. No significant differences were observed between groups (hazard ratio [HR], 1.82; 95%CI, 0.87-3.81; P=.108). When 1-year survival on the waiting list was analyzed by propensity score matching, there were no differences between groups (HR, 1.03; 95%CI, 0.36-2.97; P=.958) as shown in figure 4. No significant differences were observed in 1-year survival on the waiting list among prescription patterns (figure 1 of the supplementary data).

Figure 3.

One-year survival curves on the heart transplantation waiting list. 95%CI, 95% confidence interval; HR, Hazard Ratio; HT, heart transplantation.

(0.14MB).

Figure 4.

One-year survival curves on the heart transplantation waiting list adjusted by propensity score matching. 95%CI, 95% confidence interval; HR, Hazard Ratio; HT, heart transplantation.

(0.13MB).

A significant proportion of patients (n=203; 21.7%) had primary graft failure after HT and 167 patients (17.9%) required mechanical circulatory support. No differences between groups were found in primary graft failure (17.1% vs 23.0%; P=.068) or mechanical circulatory support implantation (18.0% vs 17.8%; P=.937). Survival after HT was similar among groups (HR, 0.79; 95%CI, 0.53-1.17; P=.238; and see figure 5) even after adjustment by propensity score matching (HR, 0,97; 95%CI, 0.60-1.56; P=.958; and see figure 6).

Figure 5.

Survival after heart transplantation. 95%CI, 95% confidence interval; HR, Hazard Ratio.

(0.15MB).

Figure 6.

Survival after heart transplantation adjusted by propensity score matching. 95%CI, 95% confidence interval; HR, Hazard Ratio.

(0.13MB).

First, this study has the limitations inherent to its observational design and retrospective collection of data on the use of levosimendan. However, all the data on the pre-HT situation and post-HT outcomes had been previously collected in the national transplantation registry and were transferred directly to our database. Because patients who received levosimendan were not as sick as patients not receiving the drug, we performed propensity score matching to compare survival between the groups. Given the retrospective nature of our data, our aim was not to compare the 2 groups, but rather to evaluate the safety of this treatment.

Second, when patients are treated with ambulatory levosimendan infusions, other intravenous treatments such as furosemide or iron may also have been prescribed and this information was not available. The follow-up during infusions tends to be closer, which may help detect HF decompensations and better adjust diuretics and HF medication.

Third, among patients who received levosimendan, 55% had a previous infusion during a prior hospitalization, which could potentially have led to selection of patients without previous adverse events.

Finally, limitations inherent to our registry data are the lack of a standardized algorithm on when and how to initiate levosimendan, the heterogeneity of prescription patterns (with no administration study protocol), and the varying lengths of levosimendan exposures among patients.