Early Experience With Percutaneous Pulmonary Blood Flow Restrictors to Allow Deferral of Norwood Surgery

Introduction

Neonates with single-ventricle anatomy are traditionally palliated with a three-stage surgical approach culminating in the Fontan operation. Stage I palliation is performed during the neonatal period and allows for unobstructed systemic blood flow and regulated pulmonary blood flow. Stage 2, the bidirectional Glenn (BDG), is typically performed at four to six months of age. However, certain clinical demographics can make patients high-risk for neonatal single-ventricle surgical palliation. Subsequently, a hybrid procedure was developed involving transcatheter ductus arteriosus stenting and bilateral surgical external pulmonary artery (PA) bands to restrict PA blood flow and avoid overcirculation and pulmonary hypertension as the patient awaits a Norwood. This deferral allows for optimization of medical comorbidities; however, this approach still requires a sternotomy and early surgery. Recent developments in single-ventricle palliation have involved off-label use of microvascular plugs (MVP) (Medtronic) as transcatheter PA flow restrictors (PAFR) to defer early cardiac surgery in neonates for medical optimization and growth.^1–3 The MVPs are modified for flow restriction by placing calibrated perforations in the plug membrane. While studies have highlighted the safety and efficacy of these devices in high-risk Stage I palliation (S1P) candidates, their role in the management of standard-risk patients requires further evaluation.^4,5 We hypothesized that PAFR can safely and effectively defer S1P in both high-risk and standard-risk neonates. Our case series describes a single-center experience with transcatheter PAFR in both standard and high-risk S1P candidates.

Materials and Methods

Flow Restrictor Modification

Microvascular plugs (Medtronic) were modified by perforating the covering using a needle and then dilating with a sheath dilator to the desired size (Figure 1). The size of the PA was measured proximally and at the takeoff of the upper lobe. If the largest PA size was <4 mm a 5Q MVP was placed, 4 to 5.5 mm a 7Q MVP, and 5.5 to 7 mm a 9Q MVP. Fenestrations were generally created with a 4F dilator in 5Q MVP and 5F dilator in 7Q or 9Q MVP. These were then deployed into branch pulmonary arteries through a transcatheter approach. Flow was documented with echocardiography and angiography post device placement, with a goal of improved systemic blood pressure, reduction in systemic oxygen saturation, and evidence of flow restriction on terminal angiography.

OPEN IN VIEWER

Figure 1.

Modified microvascular plugs (Medtronic) with perforations using a needle and then dilating with a sheath dilator to the desired size.

Results

Between December 2022 and March 2025, 12 S1P-eligible neonates underwent bilateral transcatheter PA flow restriction, with a median follow-up time of 182 days (IQR, 47-261). Nine patients had a diagnosis of hypoplastic left heart syndrome (HLHS), one patient had a diagnosis of double-outlet right ventricle, one patient had a diagnosis of double-inlet left ventricle, and one patient had a diagnosis of heterotaxy with AV septal defect and left ventricular outflow tract obstruction. Median gestational age was 37 weeks and 1 day (IQR, 34 weeks and 6 days – 38 weeks and 4 days) and median birthweight was 2.7 kg (IQR, 2.2-3.2). Nine patients were deemed high-risk by our criteria outlined above and three patients were standard-risk (Table 1).

OPEN IN VIEWER

Table 1.

Clinical Findings of Patients who Underwent Bilateral PA Flow Restrictor Placements.

Patient	Cardiac diagnosis	Gestational age	Birth weight (kg)	High-risk clinical features	Age at flow restrictor placement (d)	Duration of flow restrictors in situ (d)	Follow-up time (days)	Need for PA stent after S1P	Need for PA stent after BDG	Variance from expected post-BDG pathway	Current status
1	HLHS (MS/AS)	34w3d	2.24	Low birth weight	2	38	357	Yes	Yes	Yes	BDG
2	DORV, unbalanced right-dominant AVSD	38w2d	2.95	AV valve regurgitation	2	200	368	NA	NA	NA	Flow restrictors removed at the time of orthotopic heart transplant
3	HLHS (MA/AA)	37w2d	2.72	Highly restrictive atrial septum, IUGR, Kabuki Syndrome	2	NA	23	NA	NA	NA	Deceased prior to S1P due to inoperable neurological comorbidities
4	HLHS (MS/AS)	37w1d	2.13	Low birth weight	3	19	192	No	Yes	Yes	Deceased due to failure of BDG physiology
5	HLHS (MA/AA)	39w0d	3.69	None	4	6	286	No	No	Yes	BDG
6	DILV with CoA	39w1d	3.3	None	2	12	218	No	No	Yes	BDG
7	HLHS (MS/AS)	38w4d	3.2	Vascular Ehlers-Danlos Syndrome	2	27	252	Yes	No	Yes	BDG
8	HLHS (MS/AA)	32w3d	1.4	Low birth weight	2	NA	2	NA	NA	NA	Deceased prior to S1P due to mixed cardiogenic/septic shock
9	HLHS (MS/AA)	39w0d	2.6	Jacobsen Syndrome	3	22	172	Yes	Yes	Yes	BDG
10	HLHS (MS/AA)	34w1d	1.2	Low birth weight	4	NA	19	NA	NA	NA	Deceased prior to S1P
11	HLHS (MS/AA)	36w6d	2.5	None	7	4	55	No	NA	NA	Home awaiting BDG
12	HTX, AV septal defect, LVOTO	35w1d	3.2	HTX	34	21	68	No	NA	NA	Deceased due to sepsis and metabolic illness
Median (IQR)		37w1d (34w6d-38w4d)	2.7 (2.2-3.2)		3 (2-5)	21 (12-27)	182 (47-261)

Abbreviations: AA, aortic atresia; AS, aortic stenosis; AV, atrioventricular; AVSD, atrioventricular septal defect; BDG, bidirectional Glenn; CoA, coarctation of aorta; d, days; DILV, double inlet left ventricle; DORV, double outlet right ventricle; HLHS, hypoplastic left heart syndrome; HTX, heterotaxy; IUGR, intrauterine growth restriction; IQR, interquartile range; LVOTO, left ventricular outflow tract obstruction; MA, mitral atresia; MS, mitral stenosis; NA, not applicable; PA, pulmonary artery; S1P, stage 1 palliation; w, weeks.

There was a total of 24 MVP placed in 12 patients: 9 5Q MVP, 10 7Q MVP and 5 9Q MVP. There were two MVPs that overly restricted the PA with desaturation and poor flow documented. These devices were removed, and a second fenestration was created (patient 2 right pulmonary artery [RPA] and patient 5 left pulmonary artery [LPA]). Toward the end of the study, two fenestrations were created in patients 8, 10, and 11. The final patient had a cell removed in 9Q devices. Flow restrictor placement was performed under general anesthesia.

Flow restrictors were placed at a median age of three days (IQR, 2-5), with a median duration of placement of 21 days (IQR, 12-27) in patients who survived to the time of surgery (Table 1). Concomitant procedures included balloon atrial septostomy in six patients and patent ductus arteriosus stenting in three patients. Prostaglandin E was dosed at 0.01 to 0.05 µg/kg/min to ensure ductal patency. Three patients died prior to S1P that were not directly related to PA flow restriction. One was due to respiratory failure and neurological injury in HLHS/highly restrictive atrial septum (patient 3), another due to early mixed cardiogenic/septic shock (patient 8), and the third due to intractable junctional ectopic tachycardia in a 1.35 kg patient after balloon atrial septostomy performed two weeks after successful PA flow restriction (patient 10). Pulmonary vascular protection was adequate up to the time of death for all three patients. Patient 2 had intractable heart failure secondary to severe common AV valve regurgitation, ultimately undergoing heart transplant after 200 days with flow restrictors in place. There was intimal damage with flow restrictor removal necessitating branch pulmonary arterioplasty. This patient is doing well at home. Patient 9 had Jacobsen syndrome with thrombocytopenia and was not treated with prophylactic aspirin. Four days post placement, there was profound cyanosis, high inotrope requirement, and thrombus noted in the RPA device on computed tomography and thus underwent removal and replacement. Patient 12 was elected to undergo removal of flow restrictors in the catheterization lab prior to anticipated S1P, which resulted in an intimal injury to the left PA necessitating emergent S1P. No patients experienced flow restrictor migration or embolization.

A total of eight patients underwent S1P and currently six have undergone BDG. With respect to clinical status prior to surgery in patients who underwent S1P, four of eight patients were off vasoactive infusions aside from milrinone, one was off mechanical ventilation, five were tolerating at least half-goal enteral feeds, and five had appropriate weight gain greater than 10 g/d (Table 2). Milrinone was utilized for its inotropic effects, to decrease vascular resistance, and to augment cardiac output prior to S1P. Indications for additional vasoactive medications included hypotension and clinical evidence of low cardiac output. S1P involved a Norwood repair with a Sano shunt in six patients and a Blalock-Taussig-Thomas shunt in two patients. One patient (patient 7) required ECMO intraoperatively due to inadequate pulmonary blood flow; however, was subsequently decannulated once bilateral branch PA stents were placed on postoperative day 2. This same patient required an endarterectomy-like approach for flow restrictor removal. Three of eight patients required branch PA stenting after S1P, and three of six patients required branch PA stenting after Stage 2 BDG (Figures 2 and 3, Table 1). Indications for PA stenting included branch PA hypoplasia and thrombosis. None of the six patients had expected pathways after BDG, predominantly due to prolonged intubations and hospital length of stay (Table 1). Patient 7 required an urgent BDG in the setting of hypoxia at two months of age due to rapid progression of right ventricle-pulmonary artery (RV-PA) connection narrowing, and an early postoperative cardiac catheterization revealed left PA hypoplasia, which was subsequently stented. Patient 4 died due to failure of their BDG physiology, concomitant diastolic heart failure, and subsequent multiorgan system failure three months after their Glenn.

OPEN IN VIEWER

Figure 2.

Angiography demonstrating bilateral branch pulmonary artery hypoplasia after removal of flow restrictors.

OPEN IN VIEWER

Figure 3.

Angiography demonstrating placement of stents in bilateral branch pulmonary artery with restoration of flow.

OPEN IN VIEWER

Table 2.

Preoperative Characteristics of Flow Restrictor Patients Prior to Stage 1 Palliation.

Patient	At the time of S1P
	Off vasoactive infusions aside from milrinone	Off mechanical ventilation	Tolerating at least half-goal enteral feeds	Weight gain > 10 g/d
1	Yes	Yes	Yes	Yes
2	NA	NA	NA	NA
3	NA	NA	NA	NA
4	Yes	No	Yes	No
5	No	No	No	Yes
6	Yes	No	Yes	Yes
7	Yes	No	Yes	Yes
8	NA	NA	NA	NA
9	No	No	Yes	Yes
10	NA	NA	NA	NA
11	No	No	No	No
12	No	No	No	No

Abbreviation: S1P, stage 1 palliation.

Of note, between February 2021 and June 2023, eight patients underwent bilateral surgical PA banding with a similar principle to medically optimize patients prior to surgery. Four of these patients were S1P candidates while the others underwent biventricular repair. Among the four, one patient died prior to S1P due to vasoplegic shock. One patient died several months after S1P related to septic shock. The remaining two have undergone BDG, and both have required catheter-based interventions on their branch pulmonary arteries due to stenosis.

Comment

This study describes our experiences with transcatheter PAFR in candidates for single-ventricle palliation. Flow restrictors have been reported as a reasonable strategy to defer S1P palliation in high-risk neonates.^4,5 While the majority of our patients were clinically stable at the time of S1P, our data suggest that there are a meaningful rate of PA complications requiring additional interventions and may impact post-BDG clinical status.

Pulmonary artery flow restrictors may be a short-term solution for many patients awaiting surgical management, be it S1P or transplant.^4,7 Most patients in our cohort successfully underwent S1P, and a majority of these patients survived to stage 2 BDG. However, the quality of palliation varied significantly. There may be characteristics that predispose patients to a precarious clinical course regardless of transcatheter or surgical intervention, such as extremely low birthweight or complex medical syndromes as seen in our cohort. Even in hindsight, for these patients, the decision to avoid a primary S1P was justifiable. The use of flow restrictors in standard-risk patients was part of an evolution in practice to determine whether they could offer similar benefits to surgical PA banding, to allow deferral of Norwood surgery and to permit 2 to 4 weeks of growth and maturity. Given the meaningful rate of PA complications, it is no longer our practice to place flow restrictors in standard-risk candidates.

As a novel procedure, the ideal fenestration size is unknown, and this is seen with this cohort. With a steep learning curve, the entire field has been altering the size of the fenestration with the general trend making smaller holes due to excessive pulmonary blood flow. But as fenestrations became smaller, there were concerns of flow around the soft device, and this is seen toward the end of the cohort where a second fenestration was created (and one had a cell removed) in four of the last five patients. In the initial seven patients all had a single fenestration created, with two fenestrations placed only when problems were noted with documented poor flow. The authors are aware of plans for a multicenter registry to better standardize the size of fenestrations. It is likely that in the PA size range, those with relatively oversized devices, for example, at 4.0 mm and 5.5 mm, the device will not expand as much and therefore it will likely require larger or multiple fenestrations. Furthermore, it is important to ensure flow restrictors are optimally positioned and that adequate control of pulmonary blood flow is achieved. If this is not the case, early revision with repeat cardiac catheterization is performed. Removal is a blind process where the surgeon does not see the impact of the flow restrictor on the intima. If there are issues with pulmonary blood flow after removal, then cardiac catheterization is indicated. In some cases, stenting of the prior flow restrictor site has been necessary to manage intimal disruption and its impact on pulmonary blood flow. This has affected a minority of our patients.

Our cohort had a wide range of surgical findings at the time of flow restrictor removal. There were varying degrees of flow restrictor incorporation into the intimal wall. Patient 7 required an endarterectomy-like approach for removal; however, that may be a combination of both duration of flow restrictor placement and underlying vascular Ehlers-Danlos Syndrome. Only one patient in our cohort had complications while flow restrictors were in place requiring reintervention, which is a known risk of these devices.^4,8,9 Our series has a 66% (4/6) rate for requiring PA stenting after flow restrictor removal. Three of the four patients requiring stenting had significant intimal incorporation at the time of removal. Furthermore, three of the four patients who required stenting had a duration length greater than the median reported for our cohort. In the short term, these devices may be acceptable bridges to surgical management; however, duration of placement and patient tissue characteristics may impact proximal branch PA development after flow restrictor removal. For our cohort, duration of flow restrictor removal was primarily determined by when the patient was ready to undergo surgery, whether it be S1P or transplant in those who were not candidates for S1P. Patients were progressed to surgery when they were considered medically optimized with regard to end-organ function and hemodynamic stability. Reintervention on flow restrictors was a rare occurrence in our cohort and thus, we did not routinely exchange flow restrictors every few weeks; particularly given the majority of patients had flow restrictors placed for approximately three weeks. The clinical course after BDG was highly variable as well. For some patients, such as patient 6, hypoxia was out of proportion to the size of the pulmonary arteries. It is possible that proximal flow restriction may be associated with impaired development of distal pulmonary arteriolar beds, which may cause hypoxia after superior cavopulmonary connection, despite augmentation of proximal pulmonary arteries and removal of flow restrictors.

Limitations

This study has several limitations in addition to those inherent to a retrospective analysis. Our patient cohort is small with significant heterogeneity in high-risk candidacy that may impact their course after flow restrictor placement. Our reported experience covers a period where there was a significant learning curve in the use of flow restrictors as it is still a developing technology without a purpose-built device at this time. There is significant device optimization needed that may alter its indications and extent of use. Additionally, it is unclear whether the outcomes of flow restrictors are noninferior to those of surgical bilateral PA banding. Future directions may include modifiable flow restrictors to titrate PA blood flow and additional studies to characterize optimal candidacy for and duration of flow restrictor placement.

Conclusions

Pulmonary artery flow restrictors can be a useful adjunct to delay surgical management in certain S1P candidates. However, our experience suggests a meaningful range of PA complications that may impact their ultimate clinical status after further surgical palliation. Care teams should proceed with cautious before implementing these devices broadly in their practice, particularly in standard-risk S1P candidates.