Active versus restrictive ligation strategy for patent ductus arteriosus – A retrospective two-center study of extremely preterm infants born between 22 + 0 and 25 + 6 weeks of gestational age

Background: Patent ductus arteriosus (PDA) in premature infants is associated with adverse clinical outcomes. Mode and timing of treatment are still controversial. Data are limited in the most extremely premature infants < 26 weeks of gestational age (GA), where clinical problems are most significant and patients are most vulnerable. Aims: To investigate whether different approaches to surgical closure of PDA in two large Swedish centers has an impact on clinical outcomes including mortality in extremely preterm infants born < 26 weeks GA. Study design: Retrospective, two-center, cohort study. Subjects: Infants born at 22 + 0 – 25 + 6 weeks GA between 2010 and 2016 at Uppsala University Children ’ s Hospital (UUCH; n = 228) and Queen Silvia Children ’ s Hospital Gothenburg (QSCHG; n = 220). Main outcome measures: Survival, bronchopulmonary dysplasia (BPD), and retinopathy of prematurity (ROP). Results: Surgical closure of PDA was more common and performed earlier at QSCHG (50 % vs 16 %; median age 11 vs 44 days; p < 0.01). Survival was similar in both centres. There was a higher incidence of severe BPD and longer duration of mechanical ventilation at UUCH ( p < 0.01). There was a higher incidence of ROP, IVH and sepsis at QSCH ( p < 0.05, p < 0.01 and p < 0.01). A sub-group analysis matching all surgically treated infants at QSCHG with infants at UUCH with the same GA showed similar results as the total cohort. Conclusion: Earlier and higher rate of surgical PDA closure in this cohort of extremely preterms born < 26 weeks GA did not impact mortality but was associated with lower rates of severe BPD and higher rates of severe ROP.

Surgical ligation of the ductus arteriosus was the first treatment available, but has been replaced as first-choice since the introduction of pharmacological therapy.Cyclooxygenase inhibitors are still the most commonly used medical treatment [8][9][10][11].More recently paracetamol Abbreviations: BPD, bronchopulmonary dysplasia; GA, gestational age; hs, hemodynamically significant; IQR, interquartile range; IVH, intraventricular hemorrhage; n, number; N, total number; NEC, nectrotizing enterocolitis; PDA, patent ductus arteriosus; PVL, periventricular leucomalacia; QSCHG, Queen Silvia Children's Hospital Gothenburg; ROP, retinopathy of prematurity; UUCH, Uppsala University Children's Hospital.has become another option with a proposed more favorable profile regarding side-effects although larger studies are needed to evaluate its efficacy and safety especially in extremely premature infants [12] Surgical ligation may still be indicated with refractory hsPDA or with contraindications for pharmacological treatment.Despite extensive research there is still no consensus on optimal treatment for hsPDA, as both the PDA itself as well as active treatment strategies for PDA have been associated with adverse outcomes and many preterm infants have spontaneous closure of a hsPDA without treatment [13].Given that the interventions themselves might have adverse effects, a more conservative approach towards PDA treatment has been proposed more recently [14][15][16].
The risk for developing hsPDA increases with lower gestational age (GA) at birth, and likewise does the risks to have contraindications for, or fail to respond to, pharmacological treatment.Consequently, surgical closure might be considered more often among the most preterm infants, however the majority of studies on surgical PDA closure are performed on more mature infants, and less is known about surgical treatment and outcomes in the most extremely premature infants, who have the highest morbidity and mortality rates [17,18].
The aim of this study was to compare treatment of PDA and outcomes in infants born at <26 weeks GA in two Swedish centers with different approaches to surgical PDA treatment: Uppsala University Children's Hospital (UUCH), Uppsala, and Queen Silvia Children's Hospital Gothenburg (QSCHG).During the study period UUCH and QSCHG had respectively the lowest and the highest rates of surgical PDA closure in extremely preterm infants in Sweden.They both serve as main tertiarycare neonatal centers in regions with comparable demographic populations and have similar neonatal intensive care unit (NICU) capacities.The main outcomes were mortality, bronchopulmonary dysplasia (BPD), and retinopathy of prematurity (ROP).Other morbidities assessed were NEC, IVH, sepsis and periventricular leucomalacia (PVL).

Subjects
The study cohort included all infants born during 2010-2016 with a GA of 22 + 0 to 25 + 6 weeks, at the UUCH and QSCHG or transported to these neonatal intensive care units directly after birth.By searching the Swedish Neonatal Quality Registry for Neonatal Care (SNQ), 453 liveborn infants were identified, of which five were excluded due to major congenital malformations (cardiac defects and chromosomal abnormalities).In total, 228 infants at UUCH and 220 infants at QSCHG were included.For all patient characteristics data were retrieved from medical records completed and cross-checked with data from SNQ.The study was approved by the Regional Ethical Review Board (reference number 933-17).

Total cohort, survivors at 36 weeks GA, and matched sub-group analysis
The cohorts from the two different hospitals were compared with regard to neonatal characteristics.Outcomes were investigated in the total cohort and in survivors at 36 weeks GA.To analyze more specifically outcomes in all surgically treated infants at QSCHG (n = 111), a sub-group comparison was made where each infant at QSCHG that had had surgical PDA closure was matched with an infant at UUCH (n = 111) with the same GA at birth (±1 day), and that had lived to at least the same postnatal age at which the QSCHG infant had surgery (median 11 days; IQR 6-18 days).If more than one infant at UUCH matched equally well with regard to GA, then the one with the nearest birth date was chosen.The selection of matched patients at UUCH was made blindly as to their other outcomes and comparisons of the matched groups were made in the next instance using the same parameters and methods as in the total cohort.

Definitions
GA was estimated using routine prenatal ultrasound.Survival was defined as survival to discharge.BPD was graded according to Jobe et Bancalari [19].IVH was graded according to Papile [20].Necrotizing enterocolitis (NEC) was diagnosed according to Bell's criteria [21].PVL was diagnosed using ultrasound and/or magnetic resonance imaging.ROP was classified and staged according to The International Classification of Retinopathy of Prematurity guidelines 2005 [22], and ROP treatment included laser photocoagulation or injection with anti-vascular endothelial growth factor, following the Early Treatment of Retinopathy of Prematurity guidelines [23].Sepsis was defined as clinical signs of sepsis with elevated CRP >20 and/or blood-culture positivity.Patient data were collected from medical records including the mothers' charts for obstetric characteristics and from SNQ.

PDA diagnosis and treatment
All infants underwent at least one routine echocardiography examination performed by a pediatric cardiologist during the first days of life.Additional examinations were performed the following days depending on echocardiographic findings and clinical symptoms.Echocardiographic measurements were interpreted in accordance with The Swedish Pediatric Society's guidelines for PDA diagnostics, where a predominantly left-to right shunt and at least one of the following findings indicated hsPDA: 1) ductal diameter ≥ 1.5 mm; 2) ductal diameter > % of a pulmonary artery diameter; 3) ratio of left atrial diameter/aortal diameter ≥ 1.5; 4) absent or reversed flow in the descending aorta during diastole.Pharmacological treatment included intravenous administration of ibuprofen (initially 10 mg/kg followed by 2 doses of mg/kg with 24 h interval) or indomethacin (initially 0.2 mg/kg followed by 2 more doses with 12 and 24 h interval).General contraindications for pharmacological treatment were IVH grade II-IV, suspected NEC, and renal impairment (defined as elevated serum creatinine or oliguria/ anuria).Surgical PDA closure was performed with clip or suture application.

Statistics
Data are presented as number and percentage, mean and standard deviation, or median and interquartile range (IQR).Group comparisons were made using Pearson's chi-squared test and Fisher's exact test for categorical values, and two-tailed Student's t-test or Mann-Whitney U test for continuous variables.A p-value <0.05 was considered significant.Denominators were adjusted for missing data.Statistical analyses were performed using Stata software, version 16.1 (StataCorp, College Station, TX, USA).

Total cohort and survivors at 36 weeks GA
Neonatal characteristics are displayed in Table 1.Patients at UUCH had lower GA and birthweight (24 + 3 vs 24 + 5 GA, p < 0.05; 661 vs 705 g, p < 0.01).UUCH had more outborn infants and cesarean deliveries.Survival to discharge was similar between the two centers: 156/ 228 (68 %) at UUCH and 159/220 (72 %) at QSCHG.At QSCHG patients died within 24 h, of which 9 did not receive active treatment at birth.At UUCH, all infants born received active treatment at birth, patients died within 24 h.Of all infants born, 37 (16 %) at UUCH and 111 (50 %) at QSCHG were surgically treated for PDA, p < 0.01.Median age at surgery was 44 days (IQR 35-61) at UUCH and 11 days (IQR 6-18) at QSCHG (p < 0.01).The rates of surgical PDA closure did not change over time in either of the hospitals.Of the surgically treated patients, 44 % (QSCHG) and 41 % (UUCH) respectively did not receive pharmacological treatment for PDA because of contraindications.At QSCHG, one infant died due to cardiac tamponade as a complication of central line insertion before surgery, and one infant at each site suffered vocal cord paralysis.No other post-operative complications were reported.Pharmacological treatment rates were higher in QSCHG than in UUCH, although not in the 24-25 weeks GA sub-group.The median postnatal age at treatment was three days in both groups.The majority of all pharmacologically treated patients at QSCHG was treated with ibuprofen (84 %), as well as 100 % of the pharmacologically treated patients at UUCH.Outcomes of the total cohort and survivors at 36 weeks GA are summarized in Tables 2 and 3. Infants at UUCH had higher rate of severe BPD and longer duration of mechanical ventilation, while IVH grade III-IV, severe ROP, and clinical sepsis were more common at QSCHG.All severe IVH cases at QSCHG had occurred pre-ligation.Note that a larger number of infants with IVH III-IV at QSCHG survived (76 % vs 29 %, p < 0.01).

Matched groups
Neonatal characteristics for the matched groups and outcomes for survivors in these groups are shown in Tables 4 and 5 respectively.Twenty-one (19 %) of the infants at UUCH were surgically treated, and the median age at treatment was similar to that of the full cohort with 44 (IQR 39-54) days.Unlike in the non-matched cohort there was no difference in the frequency of pharmacological treatment in the matched group.Median postnatal age at pharmacological treatment was similar as in the total cohort.Severe BPD was more common at UUCH in general and specifically in infants born at 24-25 weeks of GA, but the duration of mechanical ventilation was similar in this sub-group.Similar to the total cohort comparison, IVH III-IV, severe ROP and clinical sepsis were more common at QSCHG.

Discussion
This retrospective study analyzes the outcome in 428 extreme premature infants born before 26 weeks of gestation in two centres in Sweden with a highly different approach to surgical closure is presented.Survival was similar in the two centers but a low surgical closure rate (UUCH) was associated with a higher incidence in severe BPD while a high surgical closure rate (QSCHG) was associated with a higher incidence of severe ROP.The rates of NEC and PVL were similar.Furthermore, a low surgical closure rate was associated with a longer duration of mechanical ventilation, while IVH grade III-IV and sepsis were more frequent in the center with a high surgical closure rate.Results were consistent in the subcohort analyses of all survivors and in the matched subcohort.
Baseline characteristics were mainly comparable except for a small but significant difference in GA, with the patients at UUCH being on average 2 days younger.The main risk factor for developing hsPDA without spontaneous closure is low GA at birth, hence the subgroup analysis in this study using matched cohorts to minimize this particular bias, which showed comparable results as the total cohort.High surgical closure rates did not lead to better survival rates in this cohort and with surgery infants are exposed to possible risks and harmful effects of an operation.A lower rate of surgical PDA closure was in this cohort associated with higher BPD rates.Longer exposure to a hsPDA has been observed to increase the risk for severe BPD in several studies due to the prolonged exposure to increased pulmonary bloodflow [2,3,24] and each week with a hsPDA represented an added risk of BPD as calculated by Schena et al. [25] However surgical PDA closure itself has also been linked to higher BPD rates in several studies [26][27][28].
Surgical PDA closure was associated with higher rates of ROP in this study, consistent with other studies [5,26,29].Conclusive biological explanations for higher ROP rates after PDA surgery are lacking but perioperative and postoperative factors have been postulated having an effect on neurosensory impairment, including hypothermia, cardiorespiratory instability and exposure to anaesthetic drugs [30].
There are several factors complicating research on PDA surgery in preterm infants, and studies presenting results associating surgical PDA closure with adverse outcomes have been subject to debate due to possible selection bias, as the need for surgical PDA closure might indicate an overall more fragile state and predisposition for other morbidities regardless of the intervention as well as a survival bias [31][32][33][34].This study is limited mainly by the fact that the two hospitals may differ in treatment routines in other aspects than that regarding surgical PDA closure, although neonatal intensive care strategies in general adhered to the same national guidelines.Although this is inevitable with inter-hospital comparisons, it cannot be excluded that differences in other treatment routines might have had an additional effect.
Regarding BPD, differences in ventilator and medication dosing routines such as postnatal corticosteroids and diuretics could have contributed to differences in severe BPD.
Regarding ROP a different treatment routine with possible clinical impact could be UUCH having lower saturation targets up till 32 weeks of GA, as this could decrease the incidence of ROP [35,36].
In QSCHG a higher rate of IVH was observed.All severe IVH cases had occurred pre-ligation, so IVH could not be linked to surgical closure itself but other factors reflecting a higher initial active approach such as a higher caesarian section rate could have contributed to less IVH at UUCH compared to QSCHG.
The time period of the cohort was limited to the years 2010-2016.This limitation is due to the fact that after 2016 a more restrictive policy towards surgery was implemented in QSCHG, in line with the international trend towards a more conservative approach during the last years.
The later timepoint of surgical PDA closure in UUCH is also reflecting a more conservative approach in line with the international trend during the last decade.In recent studies both very early and very late surgical closure have been associated with worse outcomes, the optimal  timepoint is an ongoing challenge in the rescue treatment of refractory hsPDA.
Studies investigating optimal timing of surgical closure implied that early surgical closure (<2 to 3 weeks) might have a better short term respiratory outcome and nutritional benefits, and that delayed surgical closure should be avoided to decrease the risk of adverse outcome and long term intubation needs [34,37,38].However prophylactic surgical ligation has been associated with increased BPD as well [28], further emphasizing the need to optimize the timepoint of surgical ligation.
The focus of this study is on the most premature infants born between 22 and 25 weeks of GA.Lower GA is assessed as one of the most important factors in hsPDA and its consequences and optimal strategy is most urgent in these patients, but in most studies the most premature infants are underrepresented.A focus towards more conservative treatment might be justified especially in the patients >26 weeks of GA, while a selected group of the most extremely premature infants <26 weeks of GA might benefit from a more active treatment for hsPDA [39,25], including surgical closure when pharmacological therapy fails or is contraindicated.
The evidence regarding the impact of ligation on clinical outcomes in extreme premature infants is conflicting, adverse outcomes have been correlated with both high and low rates of surgical closure, reflecting the complex relationship between adverse outcomes related to hsPDA itself, the impact of ligation, and the possibility of selection and survival bias within cohorts.

Conclusion
Earlier and higher rate of surgical PDA closure did not have an impact on mortality but was associated with lower rates of severe BPD and higher rates of ROP in our study.Due to the retrospective character of the study and the interhospital comparisons, results should be interpreted with caution.A targeted strategy for especially the patients <26 weeks GA who are at highest risk is crucial, balancing risks of treatment versus adverse impact of a prolonged hsPDA in time, and further prospective studies concerning optimal selection and timing of rescue operation especially in the lowest GA groups are needed.

Declaration of competing interest
None.

Table 2
Total cohort: Treatment variables and outcomes.

Table 3
Survivors at 36 weeks: Treatment variables and outcomes.

Table 4
Matched subgroups a : Neonatal characteristics.
a All patients at QSCHG with surgical PDA ligation were matched with a patient at UUCH according to GA.

Table 5
Matched subgroups a survivors at 36 weeks: Treatment variables and outcomes.