Follow-up after Percutaneous Patent Ductus Arteriosus (PDA) Occlusion in Lower Weight Infants

Abstract

Objectives:

To describe longer-term outcomes for infants <6 kilograms undergoing percutaneous occlusion of the patent ductus arteriosus (PDA).

Study design:

This was a retrospective cohort study of infants <6 kg who underwent isolated percutaneous closure of the PDA at a single, tertiary center (2003-2017). Cardiopulmonary outcomes and device-related complications (eg, left pulmonary artery obstruction, LPA) were examined for differences across weight thresholds (very low weight, <3 kg, and low weight, 3-6 kg). We assessed composite measures of respiratory status during and beyond the initial hospitalization using linear mixed effects models.

Results:

In this cohort of lower weight infants, 92/106 percutaneous occlusion procedures were successful. Median age and weight at procedure were 3.0 (range 0.5 – 11.1) months and 3.7 (range 1.4 – 5.9) kg, respectively. Among infants with LPA obstruction on initial post-procedural echocardiograms (n=20, 22%), obstruction persisted through hospital discharge in 3 infants. No measured variables were associated with device-related complications. Rates of oxygenation failure (28% vs. 8%; P<0.01) and decreased left ventricular systolic function (29% vs. 5%; P < .01) were higher among very low weight than low weight infants. Pulmonary scores decreased (indicating improved respiratory status) following percutaneous PDA closure.

Conclusion:

Percutaneous PDA occlusion among lower weight infants is associated with potential longer-term improvements in respiratory health. Risks of device-related complications and adverse cardiopulmonary outcomes, particularly among VLW infants, underscore the need for continued device modification. Before widespread use, clinical trials comparing percutaneous occlusion versus alternative treatments are needed.

Percutaneous (catheter-based) occlusion of a patent ductus arteriosus (PDA) is the procedure of choice for ductal closure in adults, children, and infants ≥6 kg.^1,2 In these more mature patients, percutaneous occlusion offers several benefits over surgical PDA ligation, including fewer complications, shorter recovery times, and lower health care expenditures.^{3, 4} However, use among lower weight infants (<6 kg at time of procedure) is not widely accepted, as previous studies have excluded this population, and some manufacturer recommendations specify use in patients >6 kg.^{5, 6} Although recent investigators have characterized procedural success (feasibility) rates among lower weight infants >90%, outcomes beyond the immediate post-procedural period, including the timing and nature of device-related obstruction (e.g., left pulmonary artery obstruction) and cardiopulmonary compromise (e.g., decreased myocardial performance) are not well described.^{2, 7} Additionally, medium- and longer-term respiratory outcomes, including those beyond the hospitalization, remain largely unknown.

The primary objectives of the present study were to characterize the incidence of and risk factors for device-related obstruction and cardiopulmonary compromise among lower weight infants undergoing percutaneous PDA occlusion. A secondary objective was to compare respiratory status before and after percutaneous PDA occlusion, including respiratory outcomes beyond the initial hospitalization.

METHODS

This was a retrospective observational cohort study conducted at Nationwide Children’s Hospital (NCH; #IRB17-00518), between January 2003 and February 2017. We included lower weight infants referred for percutaneous PDA closure and with a technically successful percutaneous occlusion, defined as the patient leaving the catheterization laboratory with a device in the PDA. We included only data from infants who underwent successful device placement to describe outcomes and potential complications related to longer-term placement of the closure device. Infants referred for percutaneous PDA closure who did not undergo successful device placement (technical failures) were excluded from primary analysis, but data on their outcomes were gathered. Exclusion criteria included prior cardiac catheterization or surgery, a concurrent procedure at the time of PDA closure (e.g., atrial septal defect closure), critical congenital heart disease,⁸ and a diagnosis limiting longer-term survival (e.g., Trisomy 13).⁹ No infant was excluded on the basis of gestational age, illness severity, ductal diameter, or ductal length.

Data collected included birth weight, gestational age at birth, sex, additional cardiac defects (e.g., atrial septal defects, ventricular septal defects), diagnosis of bronchopulmonary dysplasia (BPD),¹⁰ indications for PDA closure, presence of known genetic or chromosomal syndrome (e.g., Trisomy 21), age in months at time of procedure, and weight at time of procedure (very low weight, <3 kg, low weight, 3-<6 kg).

Echocardiograms were performed in accordance with guidelines from the American Society of Echocardiography.¹¹ When feasible, measurements were obtained in triplicate and averaged. In general, infants referred for percutaneous PDA occlusion had echocardiograms prior to closure, immediately post closure (<24-48 hours), ~1-month post closure, and prior to hospital discharge. In cases without an echocardiogram at discharge, the latest reported echocardiogram was used. Additional echocardiograms were at the discretion of the attending provider.

To evaluate for device-related complications and cardiopulmonary outcomes, 840 echocardiograms (median 7 per infant, range 3-17) were independently reviewed. Definitions for adverse cardiopulmonary outcomes and device-related complications were based on previous studies.^12–16 Because composite outcomes are complicated by the magnitude of the risk of the intervention across component end points and by the relative importance of the different components, individual metrics were examined across the 2 weight thresholds.

To evaluate characteristics of the PDA prior to ductal occlusion, the shunt pattern was classified as:1) left to right; 2) bidirectional; 3) right to left. When the pattern was bidirectional, the proportion of the cardiac cycle with right-to-left shunting was measured as the time of right-to-left shunting divided by the total length of the cardiac cycle. Descending aorta diastolic flow was classified into three groups: anterograde throughout diastole, no clear direction to diastolic flow, and retrograde throughout diastole. Left ventricular internal dimension in end-diastole (LVIDd) was used as a marker of left ventricular volume loading.¹⁷

Procedural details (e.g., case duration, type of device) and adverse events (AEs) during the procedure were abstracted. Although use of heparin (timing, dose) was at the discretion of the attending interventional cardiologist, our institutional approach was to administer an initial bolus of 100 units/kg of unfractionated heparin, unless contraindicated. Additional bolus doses were provided to maintain activated clotting time >250 seconds. The definition of pulmonary arterial hypertension (PH) was based on a mean pulmonary artery pressure (mPAP) >25 mmHg with cardiac catheterization under baseline conditions.¹⁸

The primary goal of the study was to evaluate outcomes beyond the catheterization, but immediate procedural details were examined to provide a comprehensive risk/benefit profile of the procedure. When multiple device placements were attempted, only the final implant was recorded. Immediate procedural AEs were stratified according to severity level (1-5).^{19, 20} Based on previous studies suggesting differences in complication rates among infants with long tubular ducts (Type C) compared with other morphologies,⁷ Type C PDAs were compared with other PDA classifications according to standard angiographic criteria.²¹ Based on studies suggesting a higher risk profile for the Amplatzer Vascular Plug II (AVP-II; Abbott, Lake Bluff, IL) device,⁷ the AVP-II was compared with other PDA closure devices. The definitions for adverse cardiopulmonary outcomes and device related outcomes are included in Appendix A (online).

Pulmonary outcomes were quantified using a composite Pulmonary Score based on weighted clinical therapies, including type of respiratory support (mechanical ventilation, continuous positive airway pressure, or nasal cannula); need for supplemental oxygen (FiO₂); and pulmonary medications (systemic steroids, diuretics).²² Over time, lower pulmonary scores reflect improving respiratory status. Pulmonary Scores were calculated on a weekly basis 4 weeks before the procedure and up to 28 weeks post procedure. Follow-up beyond initial hospitalization included outpatient imaging (echocardiogram, ventilation/perfusion scan), clinic documentation (e.g., differences in upper/lower extremity blood pressure gradients, PH medications) and causes of death, if applicable.

STATISTICAL ANALYSES:

Variables are presented as means ± standard deviations or medians with interquartile range. Characteristics of infants who did and did not experience any device-related complication (composite), were compared using unpaired t-tests or Wilcoxon rank sum tests for continuous variables and χ² or Fisher exact tests for categorical variables; similar comparisons were made for VLW infants versus LW infants and infants with PH versus infants without PH. Additionally, the timing of the postnatal echocardiogram among infants with versus without post-ligation cardiac syndrome was assessed. In order to evaluate whether procedural weight and age modified each other’s effects on the risk of a device-related complication, we fit a multivariable logistic regression model for this composite outcome that included these factors and their interaction. To compare LVEF pre and post percutaneous occlusion, a linear mixed effects model with time point (pre-procedure, <24-48 hours post-procedure, discharge/latest follow-up), group (VLW vs. LW) and their interaction, with a random patient-level intercept, was performed. Linear mixed effects models for the continuous outcomes (Pulmonary Score) and logistic mixed models for the binary outcomes (need for mechanical ventilation, diuretic use) with the Dunnett correction for multiple comparisons. P-values less than 0.05 were considered significant.

RESULTS

Among 416 patients referred for percutaneous PDA closure at our institution, 106 lower weight infants met clinical criteria and 14 were procedural failures (Figure 1). Reasons for procedural failures are provided in Table I. Subsequent analyses are based on 92 lower weight infants that left the catheterization laboratory with a device (technical success).

Figure 1:

Flowchart of patient selection for inclusion.

Table I;

online: Failed Percutaneous PDA Closure Cases

Case	Description of Events or Issues	Age (mo)	Weight of Patient (kg)	PDA Classification	Type of Device(s), if applicable	Outcome
1	Complications requiring CPR	1.0	3.8	C	*	Surgical ligation
2	Hypoplasia of the aortic isthmus	6.0	5.1	C/D	--	Surgical ligation
3	Not amenable to percutaneous occlusion2	3.0	5.0	A	--	Surgical ligation
4	Not amenable to percutaneous occlusion2	3.0	2.3	C	6 mm AVP II	Surgical ligation
5	Not amenable to percutaneous occlusion1	1.0	3.1	C	--	Surgical ligation
6	Embolization requiring surgical retrieval	3.0	2.2	C	4 mm AVP II	Surgical ligation
7	Not amenable to percutaneous occlusion1	2.0	2.7	C	6 mm AVP II	Surgical ligation
8	Not amenable to percutaneous occlusion1	0.5	4.0	A	5/4 ADO	Surgical ligation
9	Embolization requiring retrieval	1.0	3.0	C	6 mm AVP II	Spontaneous closure4
10	Not amenable to percutaneous occlusion1	2.0	5.9	B	--	Surgical ligation
11	Not amenable to percutaneous occlusion1	3.0	2.6	C	10 mm AVP II	Surgical ligation
12	Embolization requiring surgical retrieval	4.0	4.5	C	10 mm AVP II	Surgical ligation
13	Not amenable to percutaneous occlusion3	1.0	2.6	B	5/4 ADO	Surgical ligation
14	Not amenable to percutaneous occlusion2	3.0	3.2	B	5/4 ADO	Surgical ligation

PDA= Patent ductus arteriosus, CPR = Cardiopulmonary resuscitation (e.g., chest compressions); DA = Descending Aorta; LPA = Left Pulmonary Artery

^*Attempt at device closure aborted due to acute cardiopulmonary decompensation in the catheterization laboratory requiring CPR

^--Infant taken to catheterization laboratory for attempted percutaneous occlusion, but angiography concerning for pre-existing DA or LPA obstruction prior to device placement

¹Placement of device resulted in DA gradient that prevented safe release of device

²Placement of device resulted in LPA gradient that prevented safe release of device

³Placement of device resulted in DA gradient and LPA gradient that prevented safe release of device

⁴Echocardiogram revealed ductus closed spontaneously after the attempted percutaneous occlusion

Baseline patient characteristics of the cohort are shown in Table II. Fewer than half of infants received either prophylactic indomethacin (18/92, 20%) or medical treatment (24/92, 26%) with indomethacin or ibuprofen prior to referral for percutaneous occlusion. Most (54/92, 59%) infants had a diagnosis of BPD. The primary reason for referral was respiratory insufficiency with infants receiving either mechanical ventilation (35/92, 38%) or non-invasive respiratory support with continuous positive airway pressure or nasal cannula (57/92, 62%) at the time of referral. The majority also had evidence of left ventricular volume loading (68/92, 74%). At the time of procedure, 32/92 (35%) were VLW and 4 infants weighed <2 kg, 90/92 (98%) were >28 days at the time of catheterization and 76/92 (83%) were >2 months’ postnatal age.

Table II:

Demographics

	Infants (N = 92)
Birth weight (grams)	1196 (475 – 4165)
Completed weeks of gestation1
[23 - 25 weeks], n (%)	20 (22)
[26 - 28 weeks], n (%)	22 (24)
[29 - 31 weeks], n (%)	8 (9)
[≥32 weeks], n (%)	38 (41)
Female, n (%)	50 (54)
Additional cardiac defects, n (%)	46 (50)
Atrial-septal defect (ASD)	38 (41)
Ventricular septal defect (VSD)	12 (13)
Atrioventricular septal defect (AVSD)	2 (2)
Genetic/chromosomal anomalies, n (%)	16 (17)
Age at Procedure (months)2	3.0 (0.5 – 11.1)
Weight at Procedure (kg)2	3.7 (1.4 – 5.9)

Data shown as N (%) unless otherwise stated

¹Age at birth unknown (n=4)

²Mean (Standard Deviation)

Prior to percutaneous occlusion, PDA shunt flow was left to right (81/92, 88%) or bidirectional (11/92, 12%); none were exclusively right to left. Among those with a bidirectional shunt, the mean (± SD) proportion of the cardiac cycle with right-to-left shunt was 23% (±12%). Descending aorta diastolic flow was anterograde throughout diastole (50/92, 54%), no clear direction to diastolic flow (4/92,4%), and retrograde throughout diastole (38/92, 41%).

Immediate procedural outcomes and complications are shown in Table III. Femoral arterial and venous access was obtained in all cases. Classification of ductal morphology included: Type A in 19 (21%), Type C in 40 (43%), Type E in 15 (16%), and Complex (“mixed” type ducts) in 18 (20%) of cases. At the time of PDA closure, 58 infants (63%) had evidence of PH.¹⁸ In 37 cases (40%), the procedure required re-positioning with the same device (n=20) or replacement with a larger device (n=17). No deaths were observed during the procedure.

Table III;

online: Immediate Procedural Data and Adverse Events

	Infants (N = 92)
Procedural Data
Radiation Dose (mGy)1	83.5 (16.0 – 880.0)
Contrast Dose (mL/kg)1	5.0 (2.1 – 11.1)
General Anesthesia	92 (100)
Minimal PDA Length (mm)	10.1 (4.3 – 22.0)
Minimal PDA Diameter at the Aortic Ampulla (mm)	5.2 (1.3 – 11.3)
Minimal PDA Diameter at the Pulmonary Artery (mm)	3.1 (1.0 – 7.4)
Arterial Catheter Size
2.5F	1 (1)
3F	22 (24)
3.3F	36 (39)
4F	33 (37)
Venous Catheter Size
4F	13 (14)
5F	52 (57)
6F	25 (27)
7F	3 (3)
Devices Used for PDA Closure
ADO	26 (28)
AVP I	4 (4)
AVP II	61 (66)
Flipper coil	1 (1)
Adverse Events
Severity Level 1 (equipment failure or malfunction)	1 (1)
Severity Level 2	12 (13)
Arrhythmia, self-correcting	6 (7)
Loss of pulse or limb noted to be dusky (treatment not required)	4 (4) 1 (1)
Excessive blood loss (red blood cell transfusion not required)	1 (1)
Hematoma noted at access site
Severity Level 3	25 (27)
Required red blood cell transfusion in catheterization lab	18 (20)
Arrhythmia requiring intervention (medical)	1 (1)
Loss of pulse in limb requiring intervention (heparin)	4 (4)
Embolization/malposition requiring retrieval of device after release	2 (2)
Severity Level 4	5 (5)
Hypotension requiring inotropic support	2 (2)
Atrioventricular block requiring intervention	1 (1)
Respiratory compromise due to unanticipated extubation	2 (2)
Severity Level 5 (Death or need for ECMO)	0 (0)

Data shown as median (range) and N (% of cohort)

¹Data not available (n=11)

ADO = Amplatzer Ductal Occluder; AVP = Amplatzer Vascular Plug; ECMO = Extracorporeal membrane oxygenation

Within 24 hours after the procedure 24/92 infants (26%) received treatment for hypotension (Table IV). Among infants with oxygenation failure, 3 were started on inhaled nitric oxide (iNO). The decreases in LVEFs immediately post-procedure were greater among VLW infants (11.4%) than in LW infants (5.4%, P<0.01; Figure 2). Among infants (n=12/92, 13%) with evidence of decreased left ventricular systolic, the majority were mild (n=11); one infant weighing 2.6 kg at the time of percutaneous occlusion had a moderate decrease in LVEF (pre-procedure value, 64%; immediate post-occlusion value, 34%) this infant received epinephrine to augment cardiac performance. The timing of the immediate post-procedure echocardiogram included: 0- <12 hours in 19 (21%), 12- <24 hours in 62 (67%), 24- <36 hours in 4 (4%), 36- <48 hours in 4 (4%), >48 hours in 3 (3%). We observed no differences in the timing (mean ± SD) of the post-operative echocardiogram among infants with PLCS versus infants without PLCS (19 ± 4 hours vs. 23 hours ± 4; P=0.17).

Table IV:

Device-related complications and cardiopulmonary outcomes

	Total (n=92)	<3 kg (n=32)	≥3 kg (n=60)	P-value
Cardiopulmonary Outcomes
Treatment of Hypotension	24 (26)	10 (31)	14 (23)	0.46
Epinephrine	1(1)	1 (3)	1 (2)	-
Red blood cell transfusion	22 (24)	10 (31)	13 (22)	0.32
Epinephrine and red blood cell transfusion	1 (1)	1 (3)	0 (0)	-
Oxygenation failure	13 (14)	9 (28)	4 (8)	<0.01
Decreased LV systolic function*1	12 (13)	9 (29)	3 (5)	<0.01
Post-ligation cardiac syndrome (PLCS)	7 (8)	4 (13)	3 (6)	0.23
Composite adverse cardiopulmonary outcomes1	12 (13)	9 (29)	3 (5)	<0.01
Device-Related Complications
LPA obstruction2	22 (24)	8 (25)	14 (23)	1.00
Mild obstruction	21 (23)	8 (25)	13 (22)	0.80
Moderate obstruction	1 (1)	0 (0)	1 (2)	-
DA obstruction3	10 (11)	2 (6)	8 (13)	0.48
Residual shunting	2 (2)	1 (3)	1 (2)	1.00
Late embolization	0 (0)	0 (0)	0 (0)	-
Failed closure	2 (2)	1 (3)	1 (2)	1.00
Composite device-related complication4	30 (33)	10 (31)	20 (33)	1.00

LPA = Left pulmonary artery; DA = Descending aorta.

Data shown as N (%), of column, except where otherwise indicated.

Severity of LPA obstruction was based on the following peak instantaneous Doppler velocity obtained by echocardiogram: mild (≥2.0 m/s - <3.0 m/s), moderate (≥3.0 m/s)

^*All infants with decreased LV systolic function were mildly reduced (EF 41%–55%)

¹Denominators adjusted for missing data.

²Includes 2 cases of “late-onset obstruction” of LPA not observed on initial post procedure echocardiogram.

³Includes 3 cases of “late-onset obstruction” of DA not observed on initial post procedure echocardiogram.

⁴Some infants had multiple device-related complications.

Figure 2:

(A) Incidence of LPA obstruction observed over time; (B). Incidence of DA obstruction observed over time. “Late-onset” obstruction was defined as evidence of LPA or DA obstruction not observed on the initial post-procedure echocardiogram, but observed on subsequent imaging. The number of infants evaluated across the time points includes: pre-procedure (N=92), ≥1-month post-procedure (N=90), discharge or latest follow-up (N=89).

Immediately following device placement, LPA obstruction was observed on echocardiogram in 20 infants (22%). Among infants with evidence of LPA obstruction on the initial post procedural echocardiogram, persistent LPA obstruction at 1-month post closure and discharge was observed among 40% (8/20) and 15% (3/20) of infants, respectively (Figure 3, A). We observed 2 cases of mild LPA obstruction not observed on the initial post procedure echocardiogram, but observed on a later, subsequent echocardiogram 1-month post closure (“late-onset obstruction”); in both cases no evidence of LPA obstruction was observed at discharge.

Figure 3:

(A): Among VLW infants, compared with pre-procedure values (70.0%), LVEF deceased immediately post procedure (VLW infants, 58.5%, P <0.01; LW infants), but returned to baseline values at discharge or longest follow-up (69.1%, P=0.58).

(B) Among LW infants, compared with pre-procedure values (72.5%), LVEF deceased immediately post-procedure (67.1%, P <0.01), but returned to baseline values at discharge or longest follow-up (70.9%, P=0.24).

(C) Compared with pre-procedure values, the decrease in LVEF immediately postprocedure was greater among VLW infants (11.4%) than LW infants (5.4%, P <0.01).

Immediately following device placement, DA obstruction was observed on echocardiogram in 7 (8%) infants, with median maximal instantaneous gradients of 17 mmHg (16-38). Among infants with evidence of DA obstruction on the initial post procedural echocardiogram, persistent DA obstruction at 1-month post closure and discharge was observed among 4 infants at both time-points, respectively (Figure 3, B). We observed 3 cases of DA obstruction not observed on the initial post procedure echocardiogram, but observed on echocardiograms 1-month post procedure (“late-onset obstruction”); in 2 of the 3 cases of “late-onset obstruction,” no evidence of DA obstruction was observed on the echocardiogram at discharge. Among the cohort, we observed 3 infants with evidence of LPA and DA obstruction. Agreement on evidence LPA or DA device obstruction (yes/no) and severity of obstruction (mild versus moderate) was good, with κ = 0.93 and 0.85, respectively.

Two infants (2/92, 2%) met criteria for procedural failure. Both cases had evidence of persistent hemolysis secondary to the device and underwent surgical ligation of the ductus and device removal. Two infants (2/92, 2%) had evidence of residual ductal shunting by color flow Doppler on the initial post procedure echocardiogram, but this was not apparent on subsequent studies. We observed no evidence of late embolization. None of the pre-procedural variables of interest were associated with the composite outcome of a device-related complications (Table IV).

We observed no differences in adverse cardiopulmonary outcomes among infants with PH versus infants without PH (Table VI). When a multivariable model that included procedural weight (centered), procedural age (centered), and an interaction between weight and age was fit for the occurrence of a device-related complication, none of the variables predicted the composite outcome.

Table VI;

online: Comparison of adverse cardiopulmonary outcomes among infants with PH versus infants without PH

	Pulmonary Hypertension (N = 56)	No Pulmonary Hypertension (N = 36)	P-value
Treatment of Hypotension	14 (25)	10 (28)	0.96
Oxygenation Failure	9 (16)	4 (11)	0.56
Decreased LV systolic function1	7 (13)	5 (14)	1.0
Post-ligation cardiac syndrome (PLCS)	5 (9)	2 (6)	1.0
Composite adverse cardiopulmonary outcomes1	7 (13)	5 (14)	1.0

Data shown as N (%), of column, except where otherwise indicated.

¹Denominator adjusted for missing data.

Compared with pre-procedure values, the Pulmonary Score decreased over time following PDA closure (Figure 4, A, linear mixed effects model, P<0.01). Compared with pre-procedure baselines, the likelihood to receive mechanical ventilation (Figure 4, B, logistic mixed effects model, P<0.01), and the use of diuretics (Figure 4, C, logistic mixed effects model, P<0.01), decreased following PDA closure.

Figure 4:

X-axis represents time (days), with negative values denoting days prior to procedure. Y-axis designates outcomes of interest. (A) Pulmonary Score; linear mixed effects models with Dunnett’s correction for multiple comparisons over time in the Pulmonary Scores of the cohort before and after catheterization. *Adjusted P<0.01). (B) Compared with pre-procedure baseline (−2 days = 2 days before procedure) the proportion of infants on mechanical ventilation following PDA closure decreased (logistic mixed effects model, *Adjusted P<0.01). (C) Compared with pre-procedure baseline (−2 days = 2 days before procedure) the proportion of using diuretics following PDA closure decreased (logistic mixed effects model, *Adjusted P<0.01). Diuretics included furosemide, bumetanide, chlorothiazide, hydrochlorothiazide, and spironolactone.

Outpatient data were available in 81 (88%) infants with a median follow-up time of 3 years (1-8 years). Among 3 cases with persistent mild LPA obstruction at discharge, 2 had ventilation/perfusion scans at 10 and 37 months post procedure, respectively. One scan showed asymmetry normal flow split (57% to the right lung, and 43% to left lung), the other showed asymmetric split lung perfusion (77% to the right lung, 23% to the left lung). No infants have undergone re-intervention to address LPA obstruction. Among 5 cases with evidence of DA obstruction at discharge, none have had symptomatology attributed to the obstruction, with a median upper extremity/lower extremity gradient of <15 mmHg in all patients. None have undergone re-intervention to address the DA obstruction.

Among 56 infants with evidence of PH prior to the procedure,12 (21%) had evidence of PH at discharge and were followed in the outpatient cardiology clinic. Among these infants, following normalization of PH on echocardiogram, 11 have been weaned off anti-PH medications at last known follow-up (median 3 years, range 1-12 years); one infant remains on sildenafil at 52 months’ postnatal age.¹⁸ We observed 4 deaths following initial hospital discharge, none of which were attributed to the percutaneous occlusion: presumed sepsis (n=2), liver dysfunction with associated coagulopathy due to macrophage activation syndrome (n=1), and respiratory arrest (n=1).

DISCUSSION

The main finding of our study is that percutaneous PDA occlusion in lower weight infants may offer longer-term respiratory benefits,²³ but also have risks of adverse cardiopulmonary outcomes and device-related complications. Although the present study provides insights into the risk/benefit profile of percutaneous PDA occlusion in lower weight infants, the absence of a direct comparison group of infants with similar PDA dynamics and risk factors who did not undergo percutaneous PDA closure limit the interpretability of the current findings.

Consistent with previous studies, 10-15% of lower weight infants referred for percutaneous PDA closure were not closed in the catheterization suite and were referred for surgical ligation.² Although not used in the present study, novel devices that address differences in ductal morphology among lower weight infants compared with more mature counterparts are now available.²³ The US Food and Drug Administration recently approved the Amplatzer Piccolo Occluder; a device manufactured with the purpose of closing the PDA among infants with weights >700 grams.²⁴ These modifications, as well as refinements in procedural techniques and increasing operator experience, are likely to change risk/benefit profiles over time.^{15, 25, 26, 27}

Similar to surgical PDA ligation, age and weight at the time of procedure may be a risk factor for cardiopulmonary compromise following percutaneous occlusion.^{18, 28} Irrespective of the method of ductal closure (percutaneous, surgery), immaturity of the myocardium and underdeveloped adaptive mechanisms to overcome the increase in LV afterload, particularly in VLW infants, may contribute to decreased cardiac performance following device placement.^{29, 30} Although rates of post-ligation cardiac syndrome in our cohort were lower than those reported among infants undergoing surgical ligation (31%-35%), those comparisons can be misleading.^{12, 13} Teixeira et al showed that rates of adverse cardiopulmonary outcomes following surgical ligation markedly decrease beyond 28 postnatal days.¹² As 98% of infants in our cohort were >28 days at the time of percutaneous PDA occlusion, lower rates of post-ligation cardiac syndrome are not surprising. Additionally, although the risks of adverse cardiopulmonary outcomes are usually present in the first hours (<12 hours) following ductal closure^{30, 31}, most infants in our cohort had an echocardiogram 12-24 hours post-procedure. In the absence of direct comparison among lower weight infants warranting PDA closure, the central question of whether surgery or percutaneous occlusion is the preferred treatment remains unanswered.² Previous reports of an association between surgical PDA ligation and adverse neonatal outcomes failed to adjust for a number of confounders (survival bias, residual bias due to confounding by indication).^{29, 32} Thus, surgery remains a reasonable option for a subgroup of higher-risk infants necessitating ductal closure.

We offer a number of explanations for higher rates of device-related obstruction than those reported previously.²⁵ Lack of standardized definitions for device-related obstruction is acknowledged,¹¹ wherein we included even mild (>2 m/sec) cases of LPA obstruction. Interestingly, our rate of LPA obstruction at longest follow-up (3%) is consistent with previous investigators.¹⁵ The timing and nature of surveillance following device-closure are not reported consistently in the literature.^{15, 16} Although post-closure surveillance in our cohort was comprehensive, including independent review of a large number of echocardiograms, there are known limitations of echocardiography in the accurate assessment of device-related obstruction.¹⁶ Although our observation of “late-onset obstruction” among a subgroup of infants may reflect changes over time in device positioning, a more likely explanation is that the obstruction was present, but not detected, on the initial echocardiogram following device closure. Because lung disease and mechanical ventilation limit acoustic windows, more robust surveillance tools (ventilation/perfusion scans) may be useful.¹⁶ Similar to previous studies,^{14, 16} a number of patients that went on to receive surgery (failed percutaneous occlusion) had pre-existing stenosis (pre-closure gradients) in the LPA and DA; this is an important consideration in evaluating the feasibility and risk/benefit profile of the procedure.

This study has several limitations. We acknowledge the risk of referral bias, as the study was conducted at a large, pediatric academic center. Baseline characteristics (BPD, genetic/chromosomal abnormalities), exposure to ductal shunting (flow and pressure related vascular remodeling), and indications for percutaneous occlusion (respiratory insufficiency), may explain observed rates of PH. As this was a retrospective study of eligible patients treated at a single institution, no a-priori power and sample size calculations were performed. The primary reason and timing of referral for PDA closure varied markedly. Despite well-defined inclusion and exclusion criteria, half of infants in the present study had additional cardiac defects. Similar to previous studies, we chose LVEF as a pragmatic cardiac marker of cardiac performance;²⁵ however, LVEF represents only a single echocardiographic measure with limitations (preload and afterload-dependent, moderate reproducibility). Thresholds to provide treatment for hypotension were not standardized. Although mechanisms remain unknown, growth and remodeling of the vasculature may have contributed to the decreases we observed in device-related obstruction over time.³³ We made efforts to adjudicate outcomes independently, but complications may be tied to a number of patient-specific variables. Observed improvements in longer-term respiratory status following percutaneous occlusion cannot be solely attributed to the procedure and may be the consequence of unknown factors (e.g., nutrition, linear growth).

Percutaneous ductal closure may offer respiratory benefits in high-risk infants, but rates of adverse cardiopulmonary outcomes in the immediate post-procedural period and device-related complications underscore the need for continued device modifications for lower weight infants. Once the optimal device for percutaneous PDA occlusion in this subgroup of infants is identified, comparative studies between percutaneous PDA occlusion versus alternative treatment strategies will be necessary to guide the practice of evidence-based medicine.

Table V;

online: Associations between patient, procedural and device-related characteristics and any device-related complication

Variable	No device-related complication (N=62)	Device-related complication (N=30)	P-value
Procedural weight1	3.7 (1.1)	3.7 (1.2)	0.97
Procedural age (months)1	3.5 (2.0)	3.1 (1.5)	0.47
Type C PDA	25 (40)	15 (50)	0.50
Genetic/Chromosomal anomaly	13 (21)	3 (10)	0.25
Evidence of PH2	39 (63)	17 (57)	0.65
AVP-II Device	39 (63)	22 (73)	0.36

Values shown are N (%) unless otherwise stated

¹Mean (Standard Deviation)

²As determined by hemodynamic catheterization during procedure

Abbreviations and Acronyms

AE

adverse event

DA

descending aorta

LPA

left pulmonary artery

PDA

patent ductus arteriosus

PLCS

post-ligation cardiac syndrome

VLW

<3 kg at time of procedure

LW

3 - <6 kg at time of procedure

Online Appendix A: Adverse Cardiopulmonary Outcomes Definitions

Treatment of hypotension:

initiation of a new inotropic agent (e.g., epinephrine) or an increase of inotropic support by >20% of the pre-procedural dose for at least 1 hour within 24 hours of catheterization. A red blood cell (RBC) transfusion was recorded, if provided for hypotension within 24 hours of the procedure. The decision to initiate treatment for hypotension was at the discretion of the attending providers.

Oxygenation failure:

evidence of an absolute increase of at least 20% in the fraction of inspired oxygen or mean airway pressure compared with the pre-catheterization (baseline) values, for >24 hours following the catheterization.

Decreased left ventricular systolic function:

In the absence of a standardized definition, an absolute decrease of ≥10% in left ventricular ejection fraction (LVEF) on the immediate (~24-48) post-catheterization ECHO compared to pre-catheterization values and evidence of LVEF <55%; severity of LVEF was based on the following: mildly reduced [EF 41%–55%] or moderately reduced [EF 31%–40%].¹ LVEF was obtained using Modified Simpson’s method, M-mode tracings (parasternal short-axis view at the level of the papillary muscles), or both.²

Post-ligation cardiac syndrome (PLCS):

the combination of both treatment for hypotension and oxygenation failure in the absence of any procedurally related etiology (e.g., hypovolemia).³

Composite adverse cardiopulmonary outcome:

decreased left ventricular systolic function and either treatment of hypotension or oxygenation failure.

Device-related Complications, Definitions: To evaluate for pre-existing obstruction, echocardiogram images were reviewed prior to device placement.

Left Pulmonary artery (LPA) obstruction:

In the absence of a standardized definition, and consistent with previous studies,^4–6 the severity of device-related LPA obstruction was based on the following peak instantaneous Doppler velocity obtained by ECHO: 1) mild (≥2.0 m/s - <3.0 m/s), 2) moderate (≥3.0 m/s). Two-dimensional imaging and color/spectral Doppler were reviewed for evidence of device-obstruction.

Descending aorta (DA) obstruction:

Consistent with previous studies,⁷ device-related DA obstruction was defined as an echocardiogram peak instantaneous gradient (mmHg) >15 mmHg or evidence of persistent antegrade diastolic flow on spectral Doppler (“obstructed pattern”).⁷ “Late-onset” obstruction was defined as evidence of LPA or DA obstruction not observed on initial post procedure echocardiogram, but on subsequent imaging.

Residual shunting:

Echocardiogram evidence of persistent ductal shunting following device placement at hospital discharge.

Late embolizations:

Cases in which the device embolized after the infant left the catheterization suite. Cases in which a device embolized during the procedure, but was retrieved percutaneously and subsequently closed during the same procedure, was considered a technical success. This was also listed as an immediate procedural AE.^{8, 9}

Procedural failures:

Cases in which the device was placed and the infant left the catheterization suite, but subsequently required surgical or percutaneous removal at a later time due to either malposition or embolization.^{8, 9}

Composite device-related complication:

A primary outcome of the study was a composite of any device-related obstruction (LPA obstruction, DA obstruction), residual shunting, late embolization, or procedural failure.

Adjudication of outcomes: Evidence of device-related obstruction (yes/no, severity) were independently adjudicated by 2 Pediatric Cardiologists with expertise in echocardiogram (C.S., C.C.). Disagreements between reviewers were resolved by discussion, and, if necessary, a 3^rd party was consulted.