Pulmonary-to-Systemic Arterial Shunt to treat Children with Severe Pulmonary Hypertension

Abstract

Background:

The placement of a pulmonary-to-systemic arterial shunt in children with severe pulmonary hypertension (PH) has been demonstrated, in relatively small studies, to be an effective palliation for their disease.

Objectives:

To expand upon these earlier findings utilizing an international registry for children with PH who have undergone a shunt procedure.

Methods:

Retrospective data from 110 children with PH who underwent a shunt procedure collected from 13 institutions in Europe and the US.

Results:

Seventeen children died in-hospital post-procedure (15%). Of the 93 children successfully discharged home, 18 subsequently died or underwent lung transplantation (20%), mean follow-up 3.1 years (range: 25 days-17 years). The overall 1- and 5-year freedom from death or transplant were 77% and 58%, and 92% and 68% for those discharged home. Children discharged home had significantly improved World Health Organization (WHO) functional class (p<0.001), 6-minute walk distances (p=0.047) and lower brain natriuretic peptide levels (p<0.001). Post-procedure, 59% of children were weaned completely from their prostacyclin infusion (p<0.001). Pre-procedural risk factors for dying in-hospital post-procedure included ICU admission (Hazard ratio (HR) 3.2, p=0.02), mechanical ventilation (HR 8.3, p<0.001) and extracorporeal membrane oxygenation (HR 10.7, p<0.001).

Conclusions:

A pulmonary-to-systemic arterial shunt can provide a child with severe PH significant clinical improvement that is both durable and potentially free from continuous prostacyclin infusion. Five-year survival is comparable to children undergoing lung transplantation for PH. Children with severely decompensated disease requiring aggressive intensive care are not good candidates for the shunt procedure.

Condensed Abstract:

This study is the largest assessment of using a pulmonary-to-systemic arterial shunt to treat children with severe PH. Children who survive the procedure have improved clinical measures with less utilization of prostacyclin infusions. Five-year freedom from death or lung transplant is comparable to the 5-year mortality in children undergoing lung transplantation for PH. Children requiring significant intensive care support prior to their shunt procedure had a significantly higher risk of death or lung transplantation post-procedure.

Introduction

Over the last twenty years, with improved medical therapies, specialized training and increased awareness, survival of children with pulmonary hypertension (PH) has steadily improved with current 5-year survival rates of 80% or better (1). Nonetheless, children still die from their disease despite the advances over the last two decades. Similar to adults with PH, children that are World Health Organization functional class (WHO-FC) IV despite PH-specific therapy, have a median survival of less than 2 years (2,3). For children with severe PH who fail medical therapy, bilateral lung transplantation has been the ultimate palliative procedure. However, survival in pediatric lung transplantation remains relatively limited with a current 5-year survival of 64% for children with idiopathic PH and 43% for non-idiopathic PH (4). Furthermore, lung transplantation involves highly technical and expensive care that is offered at relatively few centers around the world, thus limiting its accessibility.

An alternative palliation to transplant that has been utilized since the late 1990s to treat severe, progressive PH is an atrial septostomy. In most PH patients, end-stage disease is characterized by right ventricular failure with falling systolic pressures and rising end diastolic and right atrial pressures. Having an atrial communication, therefore, could “decompress” the failing right ventricle by allowing for right-to-left atrial shunting, thereby improving left ventricular preload and in turn augmenting cardiac output (albeit with overall reduced systemic oxygen saturations). A number of small series, almost exclusively in adults, have shown that an atrial septostomy can in fact improve symptoms in someone with severe PH (5). Whether it can improve survival, however, is unclear. Very few studies have looked at the role of atrial septostomy in treating children with PH (6). One retrospective study of 20 children with PH demonstrated clinical improvement after undergoing an atrial septostomy (7). Despite limited data, atrial septostomy is included as part of PH treatment guidelines for children on maximal medical therapy who exhibit symptoms of right ventricular failure, both acute (syncope) and chronic (8,9). It may also serve as a bridge to transplantation.

In 2004, a group from Paris, France introduced a new interventional option for children with severe PH (10). Their innovative idea described treating severe PH, defined as having pulmonary artery systolic pressures higher than aortic systolic pressures, by creating a surgical anastomosis between the left pulmonary artery and the descending aorta, reprising a surgical procedure first described by Dr. Willis Potts in 1946 to treat cyanotic congenital heart disease (11) (Central Illustration). In theory, by creating an unrestrictive shunt between the pulmonary and systemic arterial systems, the afterload upon the right ventricle would be reduced from suprasystemic to systemic levels. Equalizing right and left ventricular systolic pressures recapitulates the pathophysiology seen in patients with Eisenmenger’s syndrome. In doing so, the clinicians hoped to confer upon the children the superior life expectancy purported for patients with Eisenmenger’s syndrome (12). Furthermore, because the “Potts” shunt directly reduces the systolic burden upon the right ventricle it has the potential to intervene early in a child’s disease course. This contrasts with an atrial septostomy which relies upon high diastolic pressures to be of benefit, a typical end-stage disease development. In addition, while an atrial septostomy results in global hypoxemia, the Potts shunt would still direct the highest oxygenated blood to the child’s brain. In 2014, the French group detailed their results in utilizing the Potts shunt to treat 24 children with severe pulmonary arterial hypertension (13). Nineteen of the children had a traditional Potts shunt. The other 5 had a remnant patent ductus arteriosus (PDA) opened with a stent as part of a cardiac catheterization procedure. Early mortality in the study was 12.5%. However, for survivors, with a 2.1-year median follow-up, results were encouraging with significant clinical improvement seen in the majority of children. Furthermore, this improvement was associated with a significant reduction in PH-specific medications, especially in the use of continuous prostacyclin infusions. Since these initial studies, several small series have been published describing the treatment of severe PH in children with a similar shunt procedure with similar promising results (14,15). As awareness among the pediatric PH community has increased regarding the possible efficacy of a pulmonary-to-systemic arterial shunt to treat children with severe PH, so too have questions such as safety of the procedure, who might best benefit and what are the longer-term outcomes. To address these and other questions, an international registry was established among 13 institutions to collate data on a total of 110 children who had undergone a shunt procedure for their PH. This report details the registry’s findings.

Central Illustration: Potts Shunt and Pulmonary Hypertension.

Creation of an anastomosis between the left pulmonary artery and the descending aorta (Potts shunt) in a patient with severe pulmonary hypertension and suprasystemic pulmonary artery pressures. Desaturated pulmonary blood can flow into the systemic circulation, decreasing right ventricular afterload.

Methods

The registry includes 13 institutions and was formed under the auspices of the Association for Pediatric Pulmonary Hypertension (see Supplement). All institutions had approval from their respective institutional review/ethics board with appropriate consent to supply data. Study data were collected retrospectively and managed using REDCap (Research Electronic Data Capture) electronic data capture tools hosted at Washington University, St Louis, Missouri, USA. All children undergoing a pulmonary-to-systemic arterial shunt procedure aged 20 years or less to treat PH were eligible for inclusion. Number of entries per center ranged from 1 to 36. Data reflects shunts placed from 2000 to 2020. Institutions varied in their testing for brain natriuretic peptide (BNP) levels, utilizing either BNP levels or NT-proBNP levels or both. To allow for the aggregation of data, both tests were standardized as a multiple above the upper limit of normal for that test, 100 pg/ml for BNP and 300 pg/ml for NT-proBNP. Thus, a normal value for either would be 1. Surgically created shunts between the left pulmonary artery and the descending aorta were performed via a left thoracotomy or sternotomy using either 1) a direct anastomosis, 2) an interposition tube graft or 3) a valved conduit. Shunts created in the catheterization laboratory included 1) dilation (if needed) and intravascular stenting of a PDA, and 2) transcatheter shunts involving the direct puncture between the left pulmonary artery and the descending aorta with subsequent placement of a covered stent. Data on the diameter of the shunts was not collected.

Univariate Cox proportional hazards model was used to assess risk factors for death or lung transplantation with significance considered at p-value <0.05. Multivariate Cox proportional hazards model was performed using stepwise variable election method to select for risk factors for early and late events. Because of the relatively small data set and event sizes, probability of entering and remaining in the multivariate model was set at 0.4 and 0.1 with type I error rate set at 0.1 and significance considered at <0.1. Firth method was used to adjust for the potential biases in estimating hazard ratios (HR) due to small event sizes and data sets. Hazard ratios were presented with 90% (multivariate) and 95% (univariate) confidence intervals (CI). For paired, nominal data McNemar test was used. For normally distributed, independent, continuous data 2-sample t-test was used. For non-normally distributed, continuous, paired data, the Wilcoxon signed rank test was used. A p-value less than 0.05 was considered significant. “Early” or “in-hospital” event was defined as death in a child that never left the hospital post-procedure. “Late” events were defined as death or lung transplantation. Kaplan-Meir curves were generated to show event free rates. Post-hospitalization data were taken at time of last recorded follow-up and included children who subsequently died or underwent lung transplantation.

Results

Pre-procedural assessment

A total of 110 children had data entered into the registry. Mean age at the time of the shunt was 7.7 years (range 3 weeks to 20 years), with equal gender distribution (Table 1). The majority of children, 97 (88%), had their PH classified as World Symposium on Pulmonary Hypertension (WSPH) Group-1(pulmonary arterial hypertension), 3 (3%) were classified as Group-2 (PH due to left-sided heart disease) and 10 (9%) as Group-3 (PH due to lung disease and/or hypoxemia). Twenty-seven children had what each center considered hemodynamically significant congenital heart disease (CHD) and all but two (large atrial septal defect and pulmonary vein stenosis) had reparative interventions prior to their shunt procedure. Of these 27 children, 24 were classified as having Group-1 PH and 3 as Group-2 PH. CHD lesions were: transposition of the great arteries (9), ventricular septal defect (6), PDA (4), atrial septal defect (3), Shone complex (2) and other (3).

Table 1.

Characteristics Prior to Procedure, N=110 children

Age, yrs (mean, range)		7.7 years, 3 weeks-20 years
Gender-female		56 (51%)
Presence of CHD		27 (25%)
PH Group:	Group-1	97 (88%)
	Group-2	3 (3%)
	Group-3	10 (9%)
Time from diagnosis to procedure, yrs (mean, range)		3.6 years, 0–16.7 years
WHO FC:	I	1 (1%)
	II	10 (9%)
	III	61 (55%)
	IV	38 (35%)
PH-specific medications:	PDE5i	99 (90%)
	ERA	85 (77%)
	IV/SQ PGI2	70 (64%)
	All three	66 (56%)
	None	9 (8%)
ICU admission		36 (33%)
Intravenous inotropes		20 (18%)
Mechanical ventilation		14 (13%)
ECMO		11 (10%)
Procedure type:	Surgical	74 (67%)
	Direct	47
	Conduit	27
	PDA stent	26 (24%)
	Transcatheter	10 (9%)

CHD=congenital heart disease; ECMO=extracorporeal membrane oxygenation; ERA=endothelin receptor antagonist; ICU=intensive care unit; IV/SQ PGI2=intravenous or subcutaneous infusion of prostacyclin; PDA=patent ductus arteriosus; PDE5i=phosphodiesterase type 5 inhibitor; PH=pulmonary hypertension; WHO FC=World Health Organization functional class

Median time from PH diagnosis to time of shunt procedure was 2.0 years, ranging from 0 days (three children were diagnosed with PH and underwent a shunt procedure on the same day) to 16.7 years. At the time of the shunt, 99 of the 110 (90%) children were on a phosphodiesterase 5 inhibitor, 85 (77%) were on an endothelin receptor antagonist and 70 (64%) were on either intravenous or subcutaneous infusions of prostacyclin (IV/SQ PGI₂) (17 on epoprostenol, 53 on treprostinil). Sixty-one (56%) children were on all three medications, 31 (28%) on dual-therapy and 9 (8%) on mono-therapy. Nine (8%) children were on no PH-specific medications. Usage of these medications, especially of prostacyclin infusions, depended mainly on acuity of presentation and availability. Other considerations included tolerance of side effects and quality of life decisions. At the time of their shunt, 99 (90%) of the children were considered WHO functional class (WHO-FC) III-IV. Although data were limited, suprasystemic right ventricular pressures were reported in 40 of the 47 (87%) using catheterization data and in 44 of the 57 (77%) using echocardiographic data.

At the time of their intervention, 74 (67%) of the children were outpatients while 36 (33%) were being treated in the intensive care unit (ICU). Twenty (18%) children were on intravenous inotropic medications, 14 (13%) were mechanically ventilated and 11 (10%) were on extracorporeal mechanical oxygenation (ECMO). The most common (noted in 92% of the cases) reason cited for proceeding with a shunt was worsening or insufficient improvement despite the use of PH-specific medications. Other reasons included unavailability of lung transplantation and acute decompensation requiring ICU care.

Procedural data

Seventy-four (67%) of the 110 children had a surgical shunt placed (median age 7.0 years), 26 (24%) had dilation and stenting of a PDA remnant (median age 3.5 years), while the remaining 10 (9%) had a transcatheter created shunt (median age 10.5 years) (Table 1). Of the 74 children who underwent a surgical shunt, 47 (64%) were direct anastomoses between left pulmonary artery and descending aorta (a Potts shunt) while the other 27 (36%) involved the use of a conduit: a simple tube graft (23) or valved composite graft (4). Diameter and length of the conduits were not consistently specified.

Assessment of in-hospital or early mortality

Of the 110 children who underwent the shunt procedure, 17 (15%) died during their hospitalization. The median length of time from procedure to death was 10 days, with a range from the same day of the procedure to, in a unique instance, 316 days post-procedure. Reasons for death were cardiac failure (8), both acute and progressive, respiratory failure (6), thromboembolic events (2) and bleeding (1). For the 93 children who survived to discharge, the median days on mechanical ventilation was 1 (range 0–224) with a median ICU stay of 3 days (range 0–110). The median total hospitalization stay was 9 days (range 1–223). The median upper extremity/lower extremity saturation differential at time of discharge was 10% (range 0-25).

The impact of pre-procedural variables upon in-hospital mortality was assessed using univariate Cox proportional hazards analysis. Despite the wide range in ages at the time of the shunt procedure, age did not appear to impact in-hospital mortality (HR 1.0, p=0.7) (Table 2). Gender, and time from a child’s initial diagnosis of PH to their subsequent shunt procedure did not correlate with in-hospital mortality (HR 1.2, p=0.7; HR 1.0, p=0.1 respectively). The presence of pre-existing CHD also did not significantly increase risk of early mortality (18% died, 5 of 27) (HR 1.4, p=0.6). The underlying etiology of a child’s PH did appear to influence in-hospital mortality, with 50% (5 of 10) of the children diagnosed with Group-3 PH dying (HR 5.4, p=0.002) compared to 12% (12 of 97) for Group-1 (HR 0.4, p=0.1) and none (0 of 3) for Group-2 (HR 0.04, p=0.5). Whether or not a child was receiving IV/SQ PGI₂ prior to intervention did not correlate with in-hospital mortality (HR 0.7, p=0.4).

Table 2.

Univariate Analysis of Risk Factors for Early Mortality

		Hazard ratio	95% CI	p-value
Pre-procedure data
Age at time of procedure		1.0	0.9, 1.1	0.7
Gender		1.2	0.5, 3.1	0.7
History of CHD		1.4	0.5, 3.8	0.6
Time from diagnosis to procedure		1.0	0.9,1.0	0.1
PH Group:	Group-1	0.4	0.2, 1.2	0.1
	Group-2	0.04	0, 138	0.5
	Group-3	5.4	1.9, 15	0.002
Use of IV/SQ PGI2		0.7	0.3, 1.7	0.4
ICU admission		3.2	1.2, 8.3	0.02
Intravenous inotropes		2.9	1.1, 7.9	0.04
Mechanical ventilation		8.3	3.1, 22	<0.001
ECMO		10.7	3.9, 29	<0.001
Operative data: Shunt type
Surgical		0.7	0.3, 1.8	0.4
PDA stent		0.7	0.2, 2.3	0.5
Transcatheter		3.6	1.2, 11	0.03

CHD=congenital heart disease; ECMO=extracorporeal membrane oxygenation; ICU=intensive care unit; IV/SQ PGI2=intravenous or subcutaneous infusion of prostacyclin; PDA=patent ductus arteriosus; PH=pulmonary hypertension

Pre-procedural indicators of severe disease progression, including ICU admission, (HR 3.2, p=0.02), use of intravenous inotropes (HR 2.9, p=0.04), use of mechanical ventilation (HR 8.3, p<0.001) and use of ECMO (HR 10.7, p<0.001), proved significant risk factors for in-hospital mortality. Indeed, out of the 11 children who were placed on ECMO prior to their procedure, 7 died in-hospital (64% mortality). Furthermore, data indicated that utilizing the transcatheter approach to create a shunt was also associated with an increased early mortality with 40% (4 of 10) dying post-procedure (HR 3.6, p=0.03). This compares to a 12% (3 of 26) in-hospital mortality for the PDA stented shunts (HR 0.7, p=0.5) and 14% (10 of 74) for the surgically placed shunts (HR 0.7, p=0.4). Of note, two of the four children that died after transcatheter shunt placement were on ECMO at the time of their procedure. Furthermore, all three deaths after PDA stenting were in critically ill infants (<3 months of age) with Group-3 PH. Finally,given that shunt placement ranged over a 20-year period, date of surgery was assessed for in-hospital mortality and no correlation was noted (p=0.2).

Multivariate analysis was performed using variables listed in Table 2. Because of small sample sizes we considered a p-value of 0.1 as significant (see Methods). Results showed that percutaneous shunt placement (HR 3.2, CI 1.1, 9.5, p=0.07), Group-3 PH (HR 3.3, CI 1.1, 9.8, p=0.07) and use of ECMO (HR 5.1, CI 1.9, 13.2, p=0.006) continued to be risk-factors for early mortality.

Post-hospitalization follow-up

Ninety-three of the 110 children were discharged home. Average follow up was 3.1 years with a median of 2.2 years (range 25 days to 17 years). Of the 93 children discharged from the hospital, 18 (19%) had a late event: 12 died and 6 underwent lung transplantation (median duration to event: 1.8 years, range 76 days to 12.5 years). Seven (39%) of these late events occurred within one year of the procedure. Overall freedom from death or transplant for the entire cohort of 110 children was 77% at one year and 58% at 5 years (Figure 1). Excluding the 17 in-hospital deaths, the 1- and 5-year freedom from a late event was 92% and 68% respectively (Figure 2). Reasons for late events were progressive disease with right ventricular failure (8), infection (3), hemoptysis (3), PDA stent occlusion (1), accidental tracheostomy tube dislodgement (1) and refusal to take PH medications (2).

Figure 1: Freedom from Death or Lung Transplant Post-Shunt Procedure.

Kaplan-Meier survival curve for all children post-shunt procedure (N=110). One and 5-year rates of freedom from death or lung transplant were 77% and 58% respectively. Median survival was 12.5 years.

Figure 2:

Freedom from Death or Lung Transplant Post-Shunt Procedure for Children Successfully Discharged Home.

Three measures of disease severity, WHO-FC, 6-minute walk distances and BNP/NT-proBNP levels, were compared before and after shunt placement (Table 3). Post-procedure values were taken at the time of last follow-up and included children that suffered a late event. Pre-procedure, 90% of the children were classified as WHO-FC III-IV, post-procedure this percentage fell to 22% (p<0.001). For children where pre- and post-procedure 6-minute walk distances were recorded (N=30), distances improved from a mean of 363 to 406 (median 344 to 404) meters (p=0.047). Likewise, biochemical assessment of cardiac strain using BNP and/or NT-proBNP levels also showed a significant improvement post-procedure. Pre-procedure, 50 children had a mean BNP/NT-proBNP level that was on average 11.1 times above the upper limit of normal (median 3.9). Post-procedure, the average level decreased to 4.1 times above normal (median 1.0) (p<0.001). These post-procedural improvements in clinical markers for disease were accompanied by a reduction in the overall use of IV/SQ PGI₂. Of the 61 children on prostacyclin infusion pre-procedure that were discharged home, 36 (59%) were weaned completely off the drug upon subsequent follow-up (p<0.001).

Table 3.

Clinical Parameters Pre-procedure vs Post-procedure in Children Discharged home

		Pre-procedure	Post-procedure, at time of last follow-up	p value
WHO FC (% in each category), N=93:	I	1%	26%	p<0.001
	II	11%	52%
	III	59%	15%
	IV	29%	7%
BNP/NT-proBNP levels (multiple above the ULN) (mean), N=57		11.1	4.1	p<0.001
6-minute walk distance (m) (mean), N=30		363	406	p=0.047
IV/SQ PGI 2 , (number of children using)		61	25	p<0.001

IV/SQ PGI2=intravenous or subcutaneous infusion of prostacyclin; ULN=upper limit of normal; WHO FC=World Health Organization functional class

Pre-procedural variables of age, gender and history of congenital heart disease did not prove significant for late death or transplant (HR 1.1, p=0.09; HR 1.4, p=0.5; HR 1.1, p=0.8 respectively) (Table 4). As it did with early mortality, Group-3 PH proved a risk factor for a late event (HR 4.2, p=0.03). Three of the 5 (60%) children with Group-3 PH suffered a late event compared to 14 of 85 (16%) with Group-1 PH and 1 of 3 (33%) with Group-2 PH. Type of shunt, and in particular transcatheter shunts, no longer proved a significant risk factor for a late event (HR 0.9, p=0.9). Interestingly, pre-procedural factors that impacted early survival, specifically: ICU admission, use of intravenous inotropes, use of mechanical ventilation, and use of ECMO, also carried a strong negative association with late survival or transplant (HR 4.7, p=0.002; HR 6.1, p<0.001; HR 5.9, p=0.002; HR 6.5, p=0.004 respectively). Multivariate analysis of Table 4 variables showed pre-procedural use of intravenous inotropes continued as a risk factor for a late event (HR 6.1, CI 2.7, 13.6, p<0.001). These data suggest that children who are severely compromised pre-procedure continue to carry a significant risk for a poorer late outcome.

Table 4.

Univariate Analysis of Risk Factors for Late Event (Death or Lung Transplant)

		Hazard ratio	95% CI	p-value
Pre-procedure data
Age at time of procedure		1.1	0.9, 1.2	0.09
Gender		1.4	0.6, 3.7	0.5
History of CHD		1.1	0.4, 3.2	0.8
PH Group:	Group-1	0.6	0.2, 1.6	0.3
	Group-2	0.8	0.1, 3.7	0.7
	Group-3	4.2	1.2, 14	0.03
ICU admission		4.7	1.7, 12	0.002
Intravenous inotropes		6.1	2.3, 15	<0.001
Mechanical ventilation		5.9	1.9, 18	0.002
ECMO		6.5	1.8, 23	0.004
Operative data: Shunt type
Surgical		0.8	0.3, 2.2	0.7
PDA stent		1.2	0.4, 3.5	0.7
Transcatheter		0.9	0.1, 6.8	0.9

CHD=congenital heart disease; ECMO=extracorporeal membrane oxygenation; ICU=intensive care unit; IV/SQ PGI2=intravenous or subcutaneous infusion of prostacyclin; PDA=patent ductus arteriosus; PH=pulmonary hypertension

Discussion

This paper represents the most comprehensive assessment to date regarding the use of a pulmonic-to-systemic arterial shunt to treat children with severe PH. The data, compiled from 13 pediatric centers in both the US and Europe, demonstrate that these inter-arterial shunts can provide durable, clinical improvement. The majority of children who were discharged home from the procedure had a significant improvement in their WHO-FC scores, 6-minute walk tests and BNP/NT-proBNP levels. Of equal significance, was that these improvements came with a number of the children weaning off their intravenous prostacyclin infusion. With a mean follow-up of 3.1 years and a range up to 17 years, the shunt procedure provided an overall 5-year freedom from death or lung transplantation of 58%. Five-year freedom from death or transplant improved to 68% in those children who were discharged home after their shunt procedure. These outcomes are comparable to the 5-year survival one sees with pediatric lung transplantation: 64% for children with idiopathic PH and 43% for children with non-idiopathic PH (4). The registry outcomes are all the more notable knowing that children who are WHO-FC IV despite PH-specific therapy, typical of many children in this study, have a median survival of <2 years (2). Thus, data from this registry suggest that in a child with severe PH, the pulmonary-to-systemic arterial shunt is a viable treatment option.

A significant finding of this study is that children who required aggressive ICU care, including intravenous inotropes, mechanical ventilation and/or ECMO support prior to their procedure had a significantly higher risk of dying post-procedure. The starkest evidence was seen in children requiring ECMO prior to their intervention, where 7 of the 11 children died post-procedure. Not only did these markers for severely decompensated disease align with in-hospital mortality, they were also associated with late death or transplant. Since the traditional Potts shunt is an open communication, it still requires the right ventricle to generate at least systemic systolic pressures to prevent substantial blood flow back into the lungs. Perhaps, therefore, it is not surprising that a child in severe cardio-respiratory failure pre-procedure continues to fail post-procedure, suggesting that their heart failure has progressed beyond the ability of the shunt to provide benefit. In clinically stable children, one can utilize a variety of tests to assess right ventricular function and thus their suitability for a shunt. However, such assessments in critically ill children are less straightforward and may explain in part why these children, particularly ones on pre-procedural ECMO, have such high mortality rates post-procedure. Treatment of children who require such aggressive ICU support remains problematic. The instability that makes these children poor shunt candidates also increases their risk for transplantation as well (16). Perhaps further development of mechanical support techniques will be able to provide this fragile population with better outcomes (17). Overall, assessment of right ventricular function prior to shunt consideration is critical for a good outcome regardless of a child’s clinical status. Current pre-procedural registry data from echocardiography, cardiac catheterization and cardiac MRI were too incomplete to provide specific guidance. Hopefully, however, with time these gaps can be filled.

Children in this study with either Group-1 (pulmonary arterial hypertension) or Group-2 (left-sided heart disease) PH benefitted from having a shunt. Furthermore, a history of significant CHD did not appear to impact outcomes. However, children whose PH was caused by lung disease and/or hypoxemia (Group-3 PH) were at high risk for a poor outcome. Of the 10 children identified as Group-3, 8 either died or underwent transplantation. Significantly, 7 of these 8 were infants (aged 1–7 months) that required intensive care at the time of their intervention. The two event-free survivors were the oldest of the group at the time of their procedure, 10 and 17 years of age. Worse outcomes in Group-3 children may reflect the fact that their disease often includes a significant component of hypoxia that is independent of pulmonary pressures. Although conclusions are tempered by the relatively few numbers, for severely affected children (and for critically ill infants in particular) with Group-3 PH, lung transplantation may be the better intervention.

Dilating and stenting an existing PDA or its remnant would seem the safest way to create a pulmonary-to-systemic arterial shunt, eliminating surgical morbidity and minimizing post-procedural care. During pre-procedural assessment therefore, looking for a possible PDA remnant would not be an unreasonable consideration, especially in younger children. While offering similar benefits as PDA stenting but potentially to a wider patient population, the transcatheter approach to creating a shunt has obvious appeal. However, the relatively high mortality detailed here and elsewhere argue for improved refinement before this technique gains widespread acceptance (18,19). The most common technique for shunt creation in this study was a traditional Potts shunt with direct anastomosis of the left pulmonary artery and descending aorta via a left thoracotomy. However, in a number of children a tube graft was used to complete the connection. While reasons were not always delineated, in some cases the need arose due to the inability to mobilize the two vessels in direct apposition, especially in older children. Four children from this study had placement of a valved conduit via a sternotomy. This relatively newer approach has the obvious advantage of preventing significant systemic-to-pulmonic flow both in systole and diastole. Its use could also widen the number of children who might benefit from a shunt. For instance, a child who at rest has systemic to even sub-systemic right sided pressures but with exertion develops suprasystemic pressures with symptoms might benefit from such a unidirectional shunt. Another advantage includes easier control of the shunt at the time of a possible lung transplantation. Nonetheless, potential loss of valve competency and/or stenosis and its efficacy in small children are serious concerns that deserve further investigation (20).

As discussed above, this study demonstrated that children on ECMO, children with Group-3 PH and children who had a transcatheter shunt procedure were all at increased risk for early mortality. Indeed, 12 (70%) of the 17 early mortalities included at least one of these three features. If children with these risk factors were excluded from the data, the overall early mortality would drop from 15% to 6%. Furthermore, this study argues that excluding critically ill children from shunt consideration would not only improve early mortality but long term outcomes as well.

Registry data argue that better outcomes, both early and late, are achieved in children who undergo shunt placement when they are in a compensated, relatively stable state. Questions then arise at what point in a child’s disease course should one consider a shunt? In a child with suprasystemic right heart pressures, does the practitioner trial all available PH-specific medications first before offering an intervention? While the registry data does not specifically address these questions, current guidelines for treating pediatric PH suggest considering a shunt only after a child fails IV/SQ PGI₂ therapy (8,9). Yet almost 40% of the children in the registry did not receive prostacyclin infusions prior to their procedure with no obvious impact upon early outcomes. Should a shunt be considered in lieu of IV/SQ PGI₂, knowing how onerous that therapy is for the child and family and how often children wean from it post-procedure? Certainly, the possibility that a shunt could preclude the need for infusion therapy would be of prime interest to patients and their families. Equally compelling is the observation that many children in this study weaned completely from their prostacyclin infusion post-shunt. Undoubtedly freedom from infusion therapy would be a major quality of life improvement for any child. However some caution in this regard is needed. Shunts are obviously not curative and do not guarantee that a child’s disease will not progress, even after initial benefit. Perhaps longer-term outcomes would be even better if infusion therapy was maintained? In addition, the registry does not address the role that relatively newer inhaled or oral prostacyclin therapies may play in post-shunt care. More long term data will be needed to address these questions.

The issue of lung transplantation after a shunt procedure was not specifically addressed with this study. While the registry data show that treating a child with severe PH with a pulmonic-to-systemic arterial shunt can prove beneficial, it still remains a palliative procedure. Even with initial benefit, a child’s underlying disease can eventually progress beyond the ability of the shunt to help. In these situations, lung transplantation could be the only remaining interventional option. Six children in the registry had a successful shunt procedure but then went on to undergo lung transplantation. Limited subsequent data was collected in these children but 5 of the 6 survived. While more data is needed to comment, risk of transplantation in these children is likely increased in part due to the development of pleural adhesions and aortopulmonary collaterals, especially to the shunted lung. Control of shunt flow at the time of transplant also has to be taken into account. Given the challenging nature of pediatric lung transplantation itself, some transplant centers, both pediatric and adult, may consider a shunt a contraindication to future lung transplantation. Until more information is gathered, before proceeding with a shunt procedure, open discussions with families about future limitations and risks regarding transplantation would be important in helping them make informed decisions.

Study Limitations

Registry data were collected in a retrospective manner. Like all registries, data accuracy is dependent upon the multiple sites entering the information. Most assessments discussed are relatively objective, reducing informational bias. Overall, the relatively small number of children within the study and of observed events may limit the generalization of some of the findings. Data reflects shunt procedures performed over a 20-year time period. Changes in techniques, therapies and treatment strategies over that time period could affect outcomes that may not be elucidated in this study. Of note, however, year of intervention did not correlate with early, inhouse mortality (p=0.2).

Clinical Perspectives

Competency in Patient Care and Procedural Skills: Creation of a pulmonary-to-systemic arterial shunt can provide sustained benefit in selected children with severe pulmonary hypertension.

Translational Outlook: Longer-term follow-up is needed to identify factors associated with better outcomes following the shunt procedure and determine the implications for later lung transplantation.

Conclusions

This study represents the largest assessment of using a pulmonary-to-systemic arterial shunt to treat children with severe PH. Children who survive the procedure have improved clinical measures with less utilization of prostacyclin infusions. Five-year freedom from death or lung transplant is comparable to 5-year mortality in children undergoing primary lung transplantation for PH. Children who required significant intensive care support prior to their shunt procedure had a significantly higher risk of death or transplant post-procedure.

Abbreviations:

PH

pulmonary hypertension

WHO-FC

World Health Organization functional class

BNP

brain natriuretic peptide

PDA

patent ductus arteriosus

CHD

congenital heart disease

PGI₂

prostacyclin

ECMO

extracorporeal membrane oxygenation