We retrospectively assessed the clinical and echocardiographic data of 21 preterm infants born at gestational age (GA) < 31 weeks who were admitted to the neonatal intensive care unit of CHA Bundang Medical Center from January 2012 to December 2013. Exclusion criteria for this study were infants with congenital heart defects except for a small patent foramen ovale. The proposal for this study was approved by the Institutional Review Board of CHA University Bundang Medical Center, and the requirement for patient consent was waivered because the study was a retrospective one (IRB No. 2017-08-031).

Initial echocardiograms of all infants in our study group were obtained within postnatal day 3 to assess the presence of hsPDA. An hsPDA was defined as a PDA > 2 mm in transductal diameter with an unrestrictive left-to-right shunt through the PDA and pulmonary hemorrhage, or PDA > 2 mm in diameter and cardiomegaly and pulmonary edema on chest X-ray with one of the following: respiratory failure requiring the increase in ventilator settings, or hypotension requiring vasopressors.12)

The “medication” group included infants with hsPDA who were treated with ibuprofen (intravenously, one cycle comprising an initial dose of 10 mg/kg, and two consecutive doses of 5 mg/kg each 24 hours apart) or indomethacin (intravenously, an initial dose of 0.2 mg/kg along with 0.1 mg/kg 12 hours later and 0.1 mg/kg 24 hours later) at the diagnosis of the hsPDA at postnatal day 2–3. These infants underwent follow-up echocardiography after each medication cycle until documentation of closure of the hsPDA. None of the preterm infants in this group received prophylactic treatment (medical treatment within 12 hours after birth)13) for their hsPDA.

The “no-hsPDA” group included preterm infants who did not show a PDA at initial echocardiography within postnatal day 3 or who initially had a PDA which did not meet the hsPDA criteria. Serial echocardiography was performed 2 days later, and if the PDA closed spontaneously or reduced in size, no treatment was administered.

The “ligation” group included infants who, after experiencing medical treatment failure, continued enteral feeding but cardiopulmonary compromise continued, as previously described as the “conservative approach” to PDA ligation.14)

The prevalences of grade ≥ III (ventricles filling with distension) IVH15) and grade ≥ IIA NEC according to the modified Bell's staging criteria16) were evaluated. Infants were classified as having Grade IIA NEC when systemic symptoms such as temperature instability, apnea, or bradycardia were present, intestinal symptoms such as absent bowel sounds with or without abdominal tenderness were found in addition to bright red blood from rectum, and radiological signs such as intestinal dilation, ileus, or pneumatosis intestinalis were found. BPD was defined as the requirement of supplemental oxygen at 36 weeks GA.

All echocardiograms analysed for this study were obtained at a mean of 40 weeks corrected age (term equivalent age) using commercially available ultrasound equipment (Acuson SC 2000, Siemens Medical, Mountain View, CA, USA).

The PDA diameter was measured at the pulmonary end. Maximal tricuspid regurgitation (TR) jet velocity from continuous- wave Doppler was obtained to estimate right ventricular systolic pressure derived from the modified Bernoulli equation (TR jet velocity² × 4).17)18)19) Maximal TR jet velocity, LV ejection fraction, mitral inflow Doppler velocity during early diastole (E), mitral annular tissue Doppler velocity during early diastole (E′), and the ratio of mitral E/E′ was obtained according to previous recommendations.19) The isovolumic contraction time (IVCT) and isovolumic relaxation time (IVRT) were calculated from tissue Doppler imaging,19) and the LV myocardial performance index (MPI) was calculated as follows20): LV MPI = (IVCT + IVRT)/LV systolic ejection time.

Using velocity vector imaging software (version 3.0, Siemens Medical), digital images previously obtained at 70 frames per second (fps) and stored were analyzed. Two investigators blinded to the clinical characteristics of the study population performed all offline analyses.

In the apical four-chamber view, endocardial border of the LV at the onset of the QRS wave was traced manually and subsequently tracked automatically by the velocity vector imaging software to acquire LV longitudinal peak systolic ε.21) ε curves were derived from six segments (three from the LV free wall and three from the interventricular septum), and LV longitudinal peak systolic ε was identified as the highest point of the average of the six curves (Fig. 1).21) Three cardiac cycles were analyzed and averaged. Representative examples of LV longitudinal peak systolic ε at term-equivalent age in the three groups according to treatment of hsPDA are shown in Fig. 2.

All values are expressed as mean ± standard deviation or median (minimum-maximum range), as appropriate. Data were analyzed using SPSS version 24 (IBM SPSS Statistics 24, IBM Corp., Armonk, NY, USA). Continuous data between two groups were analyzed using Student's t-test or the Mann-Whitney U test, as appropriate, and data among three groups were analyzed using the Kruskal-Wallis test. Categorical variables were analyzed using chi-square analysis. A p-value < 0.05 was considered significant. Pearson's or Spearman's correlation was used, as appropriate, to determine correlations between variables. For intraobserver variability, one observer performed an offline repeat analysis of LV longitudinal peak systolic ε in 15 infants after an interval of 4 weeks. For interobserver variability, two independent observers, blinded to the clinical data of the infants at the time of analysis, performed a separate analysis of LV longitudinal peak systolic ε in 15 infants. Intraobserver and interobserver variabilities were calculated as the mean percentage error.22)

The mean percentage error for intraobserver and interobserver variabilities in LV longitudinal peak systolic ε was 18% and 20%, respectively.

No significant correlation of LV longitudinal peak systolic ε with the duration of mechanical ventilation or diameter of the PDA was found in our study.

Myocardial ε has been reported to be a more sensitive parameter than conventional echocardiographic parameters for the detection of early myocardial dysfunction in children.28) Additionally, in neonates, ε has been utilized in myocardial functional evaluation.29)30) In our study, although LV longitudinal peak systolic ε was significantly decreased in the ligation group compared to other groups, no differences in conventional echocardiographic parameters were found among the three groups. ε can be obtained by tracking along the direction of the ventricular wall and not along the ultrasound beam; therefore, it has been reported to be less angle dependent than tissue Doppler parameters, such as mitral valve E/E′.31)32) Based on our findings, we speculate that LV longitudinal peak systolic ε would enable a sensitive detection of subtle LVSD that may be present in preterm infants treated for an hsPDA.

We speculate that one reason for the decreased LV longitudinal peak systolic ε at term-equivalent age in the ligation group compared to that in the other treatment groups in our study was the higher percentages of infants below 28 weeks GA and those with BW < 1000 g in the ligation group. A higher risk of decreased LV systolic function immediately after PDA ligation in preterm infants weighing < 1000 g26) has been reported. Altered loading conditions, such as acute increase in systemic vascular resistance,25)26) occurs immediately after PDA ligation and could cause LV dysfunction in this period.25)26) As the ligation group in our study had more premature infants < 28 weeks GA, the myocardium in the ligation group would be more immature and more intolerant to changes in afterload.33) We speculate that in our study, in the ligated group, subtle myocardial dysfunction initiated at the time of surgical ligation, which coincided with sudden changes in preload and systemic vascular resistance25)26)34) may have persisted to term-equivalent age, presenting as decreased LV longitudinal peak systolic ε in this group.

Another reason for the decreased LV longitudinal peak systolic ε at term-equivalent age in the ligation group in our study may be prolonged preload exposure due to an hsPDA at the immature myocardium. In our study, PDA ligation was performed using the “conservative approach,” indicating treatment only after failure of initial medical treatment for the hsPDA, and symptoms of pulmonary overcirculation or systemic hypoperfusion persisted.14) Additionally, in our study, after medication failed to close the hsPDA, there was a delay before ligation. A delay in surgery has been reported to lower ligationassociated morbidities,14) however, none of our patients experienced postligation cardiac syndrome23) that required milrinone treatment. On the other hand, this delay resulted in prolonged preload exposure, thereby presumably accelerating LV remodeling in the ligation group. Neonatal outcomes, namely BPD and death, have been associated with prolonged exposure to an hsPDA in preterm infants undergoing PDA ligation.35) We speculate that shortening exposure to an hsPDA may be beneficial for preserving LV systolic function at termequivalent age in preterm infants. Our results support the need for selection criteria for the early detection of preterm infants at high risk of medical treatment failure of an hsPDA.5) In our study, preterm infants in the ligation group underwent ligation at a mean of 3 weeks postnatal age. Based on our results, we speculate that earlier PDA ligation than a mean of 3 weeks postnatal age might be associated with better cardiovascular outcomes in preterm infants. Further studies involving a larger group of patients will be needed to validate this assumption.

Recently, medical and surgical treatment rates for an hsPDA in preterm infants have been declining.36) The treatment strategy for an hsPDA has been challenged by “nonintervention” approaches, namely management with ventilator manipulations and diuretics only.4) Reportedly, no benefits with regard to neonatal outcomes were found in the medical or surgical treatment group when compared to the outcomes with the nonintervention approach.4) However, in our study, when initial medical treatment was administered at 2–3 days postnatal age, LV systolic function in the medication group at termequivalent age was comparable to that in the no-hsPDA group. Additionally, as previously reported,37) the duration of mechanical ventilation or other important neonatal morbidities, such as NEC and BPD, did not differ between the medication and no-hsPDA groups in our study. Importantly, adverse reactions of indomethacin for PDA treatment have included immediate impairment in myocardial deformation with partial subsequent recovery 1 hour later, owing to alterations in coronary flow and resistance.38) However, we speculate that no persistent effects of indomethacin infusion affected LV systolic function in our study, as at term-equivalent age, no difference in LV longitudinal peak systolic ε was noted between the medication and no-hsPDA groups. Additionally, the medication group showed better preserved LV longitudinal peak systolic ε compared to the ligation group in our study, and the main difference between the two groups was the duration of ductal patency. Therefore, our results support the hypothesis that prolonged hsPDA may cause adverse outcomes.3) Based on our results, we speculate that in preterm infants with an hsPDA, early PDA ligation at less than 20 days of postnatal age in cases of medical treatment failure would be beneficial for preserving LV systolic function.

In our study, LV longitudinal peak systolic ε was significantly decreased in patients who experienced NEC compared to those who did not experience NEC. Only infants in the ligation group showed NEC. In addition to quantitative assessment of “ductal steal” as documented by the ratio of celiac artery blood flow to LV output in preterm infants with an hsPDA,39) an hsPDA has been reported to be an independent risk factor for the development of NEC, while therapy with indomethacin was not significantly associated with the risk for NEC.40) Therefore, we speculate that longer duration of an hsPDA in the ligation group might have contributed to the higher prevalence of NEC.

Our study was a retrospective observational study. Therefore, we could not establish causality. The small numbers of patients in our subgroups may have limited our statistical results. There were no preterm infants who received prophylactic indomethacin or infants who had an hsPDA but did not receive treatment for the hsPDA in our study. Therefore, we could not compare the outcome of those approaches.

We speculate that in preterm infants with an hsPDA, in cases of medical treatment failure, early PDA ligation at less than 20 days of postnatal age would be beneficial for preserving LV systolic function.