We performed a single-center, retrospective study. We included preterm infants who underwent PDA ligation with a gestational age of 23–33 weeks who were admitted to the neonatal intensive care unit between January 2017 and December 2022. We excluded infants with (1) cardiac anomalies other than patent foramen ovale (PFO) and persistent left superior vena cava, (2) multiple abnormalities or apparent clinical syndrome, and (3) chromosomal abnormalities.

The referrals for PDA ligation were triaged according to the clinical and echocardiographic findings. The following factors were indications for surgical closure of PDA: (1) further ventilatory support; (2) progressive congestive heart failure despite medical management or where cyclooxygenase inhibitors were contraindicated; (3) a transductal diameter > 1.5 mm, predominant left-to-right flow; (4) left atrial (LA) enlargement indicated by an LA/aortic diameter ratio (LA/Ao) > 1.3 or LA volume index > 1.0 ml/kg; and (5) left pulmonary artery end-diastolic velocity (LPA EDV) > 15 cm/s on echocardiography. Three-dimensional cardiac volume acquisition has been part of our standard protocol since 2017.

The indication for PDA surgery was determined by neonatologists who were unaware of the study design and were blinded to the 3D echocardiographic data. The LV and RV loading conditions and function before and after PDA ligation were evaluated by transthoracic echocardiography within 12 h before PDA ligation, within 4–8 h after PDA ligation, and between 24 and 48 h postoperatively. A power calculation was not performed because of the paucity of 3D echocardiographic data in this population.

LV and RV cardiac function in preterm infants with PDA (PDA group) before ligation was compared with that in those without PDA (non-PDA group). The inclusion criteria of the non-PDA group were as follows: (1) 14-day-old neonates with a gestational age between 23 and 28 weeks and those with a gestational age between 29 and 31weeks who required mechanical ventilation who were admitted to the neonatal intensive care unit between October 2020 and December 2022; (2) neonates who did not have PDA ligation performed; (3) PDA closure at day 14; and (4) neonates who did not have any congenital heart defects or pulmonary hypertension as indicated by tricuspid regurgitation pressure gradient >32 mmHg27 and/or a non-circular LV shape at the peak of systole. Three-dimensional cardiac volume acquisition for 14-day-old neonates with a gestational age between 23 and 28 weeks and those with a gestational age between 29 and 31weeks who required mechanical ventilation has been performed since 2020. The echocardiography dataset including 3D was acquired in all neonates with these criteria.

The preterm infants were treated according to our institutional protocols. The study was conducted in accordance with the principles contained in the Declaration of Helsinki and was approved by the institutional review board of Kanagawa Children’s Medical Center (No. 1806-07).

Data on sex, gestational age in weeks, birth weight, Apgar scores, the age (days) at PDA surgery, corrected gestational age in weeks, and body weight (BW) at the surgery day were collected from medical records. In the PDA group, treatment details, and survival or death at discharge, and additional characteristics were obtained.

All data, including baseline hemodynamics, respiratory characteristics, and echocardiograms, were obtained at the three time points mentioned above.

N-terminal pro-brain natriuretic peptide (NT-proBNP) concentrations were measured at pre-ligation and 24–48 h after surgery.

The clinical data collected included heart rate, blood pressure, oxygen saturation, the fraction of inspired oxygen, and mean airway pressure. Blood pressure was measured immediately before the echocardiography was performed. Postoperative blood pressure was measured using an arterial line at all time points in the PDA group. Blood pressure was measured with the oscillometer technique in all preterm infants in the non-PDA group and at some preoperative points in those in the PDA group without an arterial line.

At the time of the echocardiography, the respiratory severity score was calculated as the mean airway pressure (mmHg) × the fraction of inspired oxygen.28 The vasoactive–inotropic score at the initial echocardiographic examination was calculated as the dopamine dose (μg/kg/min) + the dobutamine dose (μg/kg/min) + 100 × the epinephrine dose (μg/kg/min) + 100 × the norepinephrine dose (μg/kg/min) + 10,000 × the vasopressin dose (U/kg/min) + 10 × the olprinone dose (μg/kg/min).29

Echocardiographic examinations were performed by an experienced echocardiographer (K Toyoshima). LV and RV function was evaluated using 3D echocardiography as part of our institutional protocol. Full-volume 3D datasets were acquired by the apical approach either from left-sided chest wall or subcostal using an ultrasound machine and equipment (EPIC 7 G or EPIC CVx with the X7-2 probe; Philips Healthcare, Andover, MA). The depth and sector angle were manipulated to include the entire LV or RV with a frame rate of > 40 frames/s. A full-volume scan was acquired from six R wave-triggered subvolumes to include the complete LV or RV into the 3D dataset. Six cardiac cycles in each capture were stitched together. We extracted 3D data over 6 cardiac cycles under a well-sedated level, no body motion, and no change in loading conditions.

The 3D echocardiography datasets for LV and RV were analyzed by an experienced investigator (K Toyoshima) using novel 3D echocardiography software (4D LV-Analysis version 3, 4D RV-Function version 3; TomTec Imaging Systems, Unterschleissheim, Germany). The accuracy and reproducibility for those have been validated by comparison with cardiovascular magnetic resonance.30,31,32

The LV endocardial border in the LV-focused four-chamber view was semi-automatically determined after two-point clicking of the LV apex and the center of the mitral valve annulus on the apical four-, two- and three-chamber views extracted from 3D echocardiography datasets. When required, the endocardial border was manually adjusted. The software generated time domain LV volume curves, and calculated the LV volume and LVEF (Fig. 1). LV end-diastolic volume (LVEDV), end-systolic volume (LVESV), stroke volume (LVSV) (calculated as the difference between LVEDV and LVESV), LVEF, global longitudinal strain (LVGLS), global circumferential strain (LVGCS), and torsion were automatically generated by the software (Fig. 1).

ED end-diastole, ES end-systole, EDV end-diastolic volume, ESV end-systolic volume, SV stroke volume, EF ejection fraction, SDI systolic dyssynchrony index, GLS global longitudinal strain, GCS global circumferential strain.

Three orthogonal planes and various landmarks in the apical RV-focused four-chamber view were selected to define the end-diastolic frames to obtain RV volume. According to the initial view adjustment, the program automatically supplied four chamber, sagittal, and coronal RV views, as well as RV end-diastolic volume (RVEDV), end-systolic volume (RVESV), stroke volume (RVSV) (calculated as the difference between RVEDV and RVESV), and ejection fraction (RVEF) ((Supplemental Fig. 1). Body size-dependent parameters were indexed by dividing by BW. LVCO and RVCO were calculated from the LVSV and RVSV, respectively, and heart rate at 3D volume measurements. Arterial elastance (Ea) was calculated as (0.9 × systolic blood pressure/stroke volume). End-systolic elastance (Ees) was calculated as (0.9 × systolic blood pressure / end-systolic volume (ESV).1,33 Echocardiographic Ea/Ees was also calculated to assess ventriculo-arterial coupling.1,33 We calculated Ea/Ees from ESV/ stroke volume of the LV and RV as measured by 3D echocardiography.

LA volume, LAEF, LA global longitudinal strain (LAGLS), and LA global circumferential strain (LAGCS) were calculated using the LV analysis software.

The following echocardiographic variables were measured: LV diastolic dimension (mm) and LV systolic dimension (mm) using the M-mode in the long-axis view; LA diameter (mm) and Ao diameter (mm) in the long-axis view using the leading edge method;34 LA area (cm2) and LA long-axis length (LA length, cm) in the four-chamber view; the narrowest internal diameter of the PDA (mm) using 2D echocardiography in the ductal long-axis view; the PDA flow pattern using the pulsed wave Doppler (left to right, right to left, bidirectional, none); and LPA end-diastolic velocity (edv) using the pulsed wave Doppler.35,36,37 RV function was evaluated by the fractional area change and corrected tricuspid annular plane systolic excursion (tricuspid annular plane systolic excursion/RV long diameter).38 We calculated LA volume using the single-plane area–length method in the four-chamber view with the following equation: LA volume = 0.85 × (LA area)2/(LA length) (cm3).37 The LV end-systolic wall stress was calculated using mean blood pressure measurements.39,40 Superior vena cava (SVC) flow were measured by pulsed wave Doppler as Kluckow and Evans reported.41 The mean of the maximum and minimum diameters measured from a still 2D image was determined over 5 heart cycles. The velocity time integral was calculated from the Doppler velocity tracings and averaged over 5 consecutive cardiac cycles. Heart rate was measured from the peak-to-peak intervals of the Doppler velocity time signals.

PDA ligation was performed employing standard methods under general anesthesia in intubated preterm infants using a standard technique, with intravenous infusions of fentanyl and the muscle relaxant pancuronium. A standard left posterolateral thoracotomy was performed through the third intercostal space with the patients placed in a right lateral position. A single ligation of the PDA was performed with a silk suture.

Fifteen studies were randomly selected from the PDA group to investigate intraobserver variability, and one observer (K Toyoshima) measured LV and RV volumes at three month intervals. The observer at the second measurement was blinded to the results of the first measurement. A second observer (H Aoki), who was blinded to the results of the first observer, independently analyzed these data to investigate interobserver variability. Intraobserver and interobserver variabilities were examined using the intra-class correlation coefficient (ICC) and Bland–Altman analysis.

Descriptive statistics (e.g., mean ± standard deviation, median [interquartile range]) were used to summarize the demographic or clinical data of preterm infants in the PDA and non-PDA groups. Differences between the two groups were analyzed using the unpaired t-test for continuous variables, Mann–Whitney U-test for median values, or Fisher’s exact test for categorical data. The hemodynamic, respiratory, and echocardiographic parameters were compared across the three time points using one-way analysis of variance with repeated measures.

Statistical analyses were performed with EZR (version 1.54) (Saitama Medical Center, Jichi Medical University, Saitama, Japan), which is a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria) and MedCalc (version 20; MedCalc Software Ltd., Ostend, Belgium, Belgium). A P value < 0.05 was considered significant.