Study population and ethical approval
Preterm infants (< 37 weeks of GA) treated at the neonatal intensive care unit (NICU) of the University Children´s Hospital of Bonn, Germany, during the study period of 01/2018-06/2021 were retrospectively screened for study participation. Inclusion criteria: documented use of levosimendan, echocardiographic diagnosis of AHF or PH at baseline and available echocardiographic data at baseline and 24 h after onset of levosimendan administration. Exclusion criteria: congenital heart defect with need for operative correction, severe syndromic disorder or chromosomal anomalies, palliative care after birth, and infants with a congenital diaphragmatic hernia (CDH), as levosimendan treatment was already evaluated recently in a subgroup of preterm and term infants with CDH [21].
Monitoring Data
The following hemodynamic parameters were documented at baseline and at follow-up (3, 6, 9,12, 24, and 48 h after onset of levosimendan drug infusion): systolic, diastolic, and mean arterial blood pressure (MAP), heart rate, pre- and postductal peripheral oxygen saturation (SpO2), fraction of inspired oxygen (FiO2). Arterial blood gas measurements with pH, arterial oxygen partial pressure (paO2), arterial carbon dioxide partial pressure (paCO2), and arterial lactate were documented when available at baseline, and at follow-up (3, 6, 9, 12, 24, and 48 h). For stratification of the oxygenation impairment and for infants with mechanical ventilation (MV) the Oxygenation Saturation Index (OSI; \(\frac{FiO2xMAPx100}{SpO2}\)) was calculated at baseline and follow-up (12, 24 and 48h), for infants without MV the Saturation Oxygenation Pressure Index (SOPI; \(\frac{CPAP pressure or PEEP x FiO2}{SpO2}\))[23] was calculated at the respective timepoints. The vasoactive-inotropic score (VIS) was calculated at the same timepoints according to the formula described elsewhere for estimation of cardiovascular drug support[4].
Diagnosis and Treatment of AHF/ PH
Diagnosis of AHF was based on clinical findings and echocardiographic assessment. A low cardiac output syndrome (LCOS) and AHF was suspected in presence of sinus tachycardia (> 180 bpm), arterial hypotension, low urine output, prolonged capillary refill time, and elevated lactate. Echocardiographic assessment and verification of AHF was performed by the attending physician with experience in neonatal echocardiography. The echocardiographic assessment is described below in more detail. When AHF was apparent dobutamine or milrinone were used as first line inotropes for improvement of ventricular function. In case of arterial hypotension and need for increased afterload, vasopressors (norepinephrine and vasopressin) were added to the inotropic therapy. In infants with missing improvement with the standard drug therapy and severe AHF levosimendan was implemented as a second line therapy due to the decision of the attending senior physician.
Likewise, the diagnosis of PH consisted of clinical signs of PH and the echocardiographic assessment. Possible clinical sings of PH and oxygen impairment were: FiO2 > 0.4, pre- and postductal SpO2-difference > 5%, paO2 < 60 mmHg despite oxygen therapy and ventilation support, low MAP and sinus tachycardia. The echocardiographic PH assessment is described more detailed below. For the reduction of the pulmonary vascular resistance iNO was used as primary drug therapy when PH was present. In preterm infants with moderate to severe PH intravenous continuous sildenafil was added as a second line therapy, followed by bosentan as third line therapy.
Echocardiographic assessment
For echocardiographic measurements a Philips CX50 Compact Extreme Ultrasound system with a S12-4 sector array transducer (Philips Healthcare, Best, the Netherlands) was used. All available echocardiographic data at baseline (prior to start of levosimendan administration), and at follow-up (24 h after onset of levosimendan administration) were retrospectively evaluated offline for analysis independently by two experienced neonatal echocardiographers. AHF was defined as ventricular dysfunction (a) right-ventricular dysfunction (RVD), b) left-ventricular dysfunction (LVD), and c) biventricular dysfunction (BVD) and classified as: present or not present. For the assessment of ventricular dysfunction a combined approach of quantitative and qualitative measurements was used, based on international guidelines for neonatal echocardiography [15, 24, 13]. PH was graded as mild, moderate, or severe, using the following echocardiographic parameters: a) PDA flow pattern b) intraventricular septum position, and c) tricuspid valve regurgitation. Additionally, the end-diastolic right-ventricular to left-ventricular (RV/LV) ratio was calculated in a standard four chamber view directly distal to the tricuspid and mitral annulus as a horizontal line from the endocardium of the RV and LV free wall to the endocardium of the interventricular septum.
Statistical analysis and outcome measures
Infants were divided into subgroups according to the primary outcome: response to levosimendan therapy (Responder vs. Non-Responder). A response to levosimendan treatment was defined as: echocardiographic improvement of RVD, LVD or BVD after 24 h, and/ or decrease of PH severity (≥ 1 grade) after 24h. The following parameters were defined as secondary endpoints or outcome measures: decrease in arterial lactate ≥ 20% after 24h of levosimendan administration, duration of MV, days of oxygen supply, VIS at 24h and 48 h, and in-hospital mortality.
For data analysis, SPSS version 27 (IBM Corp., Armonk, NY) was used. Continuous variables were described using median and interquartile range (IQR) and categorical variables were summarized as absolute number (n) with percentage. For comparison of continuous and non-normally distributed variables, a Wilcoxon-test or Mann-Whitney U test was performed to compare continuous variables between timepoints and subgroups (Responder vs non-Responder), as appropriate. For categorical variables the Pearson’s Chi2 test and Fisher's exact test was applied, as appropriate. Correlations between variables were evaluated by Spearman correlation coefficients. A p-value of < 0.05 was considered significant.