Extracorporeal Membrane Oxygenation in Children with Pulmonary Atresia and Intact Ventricular Septum: Mortality and Associated Outcomes

doi:10.21203/rs.3.rs-5299953/v1

Introduction

Data on outcomes of extracorporeal membrane oxygenation (ECMO) are limited in patients with pulmonary atresia intact ventricular septum (PAIVS). The objective of this study was to describe the use of ECMO and the associated outcomes in patients with PAIVS.

Methods

We retrospectively reviewed neonates with PAIVS who received ECMO between 2009–2019 in 19 US hospitals affiliated with the Collaborative Research for the Pediatric Cardiac Intensive Care Society (CoRe-PCICS). Patients who received ECMO were compared to those who did not and patients on ECMO who died were compared to those who survived by bivariate analysis and multivariable logistic regression. The predictive ability of a risk score for inpatient mortality (using beta-coefficients) was assessed by receiver operator curve analysis.

Results

Of 295 identified patients, 32 (11%) were supported with extracorporeal membrane oxygenation. Of these, 15 (46%) experienced mortality. A higher left pulmonary artery z-score (beta coefficient 0.72) and the presence of ventriculocoronary connections by cardiac catheterization (beta coefficient 1.25) was associated with an increased risk of ECMO (p-value < 0.01). The resulting risk score had an area under the curve of 0.71 (p-value 0.03) for prediction of need for ECMO.

Conclusion

In a multicenter cohort of patients with PAIVS, 11% received ECMO. Of those supported with ECMO, 46% experienced inpatient mortality. A higher left pulmonary artery z-score and the presence of ventriculocoronary connections appear to be risk factors for use of ECMO.

Congenital Heart Disease

Congenital Heart Surgery

Pulmonary Atresia Intact Ventricular Septum

Right Ventricular Dependent Coronary Circulation

Pulmonary atresia with intact ventricular septum (PAIVS) is characterized by varying degrees of right ventricular hypoplasia and coronary abnormalities. In its more severe form, right ventricular coronary dependency can result in myocardial ischemia with ventricular dysfunction, cardiac arrhythmias, and possibly cardiac arrest, sometimes requiring the use of extracorporeal membrane oxygenation (ECMO). Venous inflow for ECMO may lead to RV decompression and carries the theoretical risk of worsening of catastrophic myocardial ischemia[1]. In a single center series of patients with PAIVS, only 1 out of 8 patients that received ECMO was successfully decannulated[2]. Based on these considerations, the utility of ECMO in this population has been questioned, and even contraindicated. Patient level data from large multicenter cohorts are limited[3]. Therefore, the primary objective of this study was to determine factors associated with ECMO use in neonates with PAIVS at their first hospital admission. Secondary objectives were to describe the outcomes of those who received ECMO and to evaluate the risk factors associated with inpatient mortality.

Patient identification

We retrospectively reviewed the initial neonatal admission of all patients with PAIVS admitted in the cardiac ICU to one of the 19 participating CoRe-PCICS pediatric centers in the United States between 2009 and 2019. A list of participating institutions can be found in Supplemental Table 1. The study was approved by the institutional review boards of all participating study centers and the data coordinating center at Cincinnati Children’s Hospital Medical Center (ID: 2019 − 0708; 24 May 2019) and given the retrospective nature of the data collected, the need for informed consent was waived (IRB Protocol H-45938).

We used demographic and anatomic data (obtained from echocardiogram, cardiac catheterization, and computed tomography angiography) during the first week of life to determine association with ECMO and mortality. Inclusion criteria were patients with ECMO use at any point during the initial neonatal hospitalization, including before first intervention, intraoperatively, postoperatively and/or prior to hospital discharge. Hospital mortality was defined as death prior to hospital discharge. Exclusion criteria were presence of ventricular septal defect, evidence of blood flow across the pulmonary valve, and Ebstein’s anomaly or severely dysplastic tricuspid valve. Exclusion criteria were determined based on the initial echocardiogram report.

Variables of interest

Study relevant descriptive information was collected. Also, as one of the aims was to model the use of ECMO and associated inpatient mortality using demographic, echocardiographic, and catheterization data routinely obtained in the first week of life, variables of interest were selected accordingly. The following demographic information was obtained: presence of prenatal diagnosis, gestational age (weeks), birth weight (kg), birth height (cm), gender, ethnicity, presence of chromosomal abnormality, presence of a noncardiac anomaly.

Based on previously reported data, the following echocardiographic data were obtained from the initial echocardiogram: tricuspid valve z-score, right ventricular morphology, right ventricular size, estimated right ventricular pressure, type of pulmonary atresia, pulmonary valve annulus z-score, right pulmonary artery z-score, left pulmonary artery z-score, left ventricular ejection fraction, presence of normal right coronary artery, presence of normal left coronary artery, evidence of to-from flow in the coronary circulation, and presence of ventriculocoronary connections. All participating institutions used echocardiography-derived calculations for TV Z-score determination with the majority (17 out of 19 centers) utilizing the Boston Children’s Hospital dataset. No data imputation was done.

The following cardiac catheterization data was obtained from the initial catheterization: presence of ventriculocoronary connections, presence of coronary artery stenosis, and presence of coronary ostial atresia.

Statistical analyses

Descriptive variables were described as absolute frequency and percentage. Continuous variables were described as mean and standard deviation due to data for most variables being normally distributed.

Demographic characteristics, echocardiographic data, and cardiac catheterization were compared between those who received extracorporeal membrane oxygenation and those who did not, using chi-square or Fisher exact tests for descriptive variables and independent t-tests or Mann-Whitney-U tests for continuous variables. Variables with a p-value of less than 0.2 were entered into a backward logistic regression with need for extracorporeal membrane oxygenation as the dependent variable. A model was developed using the beta-coefficients for independent variables that were found to have significant association with the dependent variable. The developed model was used to calculate the estimated risk of extracorporeal membrane oxygenation. A receiver operator curve analysis was then conducted to determine the accuracy of this risk score to predict extracorporeal membrane oxygenation use.

Among ECMO patients, analyses compared those that did and did not experience inpatient mortality during the admission. Demographic characteristics, echocardiographic data, and cardiac catheterization data were compared using chi-square of Fisher exact analyses for descriptive variables and independent t-tests of Mann-Whitney-U tests for continuous variables. Variables with a p-value of less than 0.2 were entered into a backward logistic regression with inpatient mortality as the dependent variable. A model was developed using the beta-coefficients for independent variables that were found to have significant association with the dependent variable. The developed model was used to calculate the estimated risk of inpatient mortality on extracorporeal membrane oxygenation. A receiver operator curve analysis was then conducted to determine the accuracy of this risk score to predict inpatient mortality in patients requiring extracorporeal membrane oxygenation.

All statistical analyses were conducted using SPSS Version 23.0. A p-value of less than 0.05 was considered to be statistically significant. Any use of the word “significant” or “significantly” in the manuscript refers to statistical significance unless explicitly specified.

Univariable analyses, use of ECMO

Of the 295 patients with PAIVS in this registry, 32 (11%) received ECMO. No significant differences were noted in demographic characteristics (Table 1). Right pulmonary artery z-score and left pulmonary artery z-score were significantly higher in those who required ECMO (Table 2). Ventriculocoronary connections and coronary ostial atresia were more frequently noted by cardiac catheterization in those who received ECMO (Table 3).

Table 1. Cohort Characteristics

Variables	No ECMO (n=263)	ECMO (n=32)	p-value
*Prenatal diagnosis*	181 (68)	27 (84)	0.06
*Gestational age (weeks)*	38.0 ± 2.3	37.7 ± 2.2	0.44
*Weight (kg)*	2.9 ± 0.6	2.8 ± 0.5	0.18
*Height (cm)*	48.0 ± 4.4	47.6 ± 3.8	0.60
*Gender* *Male* *Female*	130 (49%) 133 (51%)	13 (41%) 19 (59%)	0.34
*Ethnicity* *Hispanic* *Not Hispanic* *Not reported/unknown*	48 (18%) 206 (78%) 9 (4%)	11 (34%) 21 (66%) 0 (0%)	0.06
*Chromosomal abnormality*	31 (12%)	1 (3%)	0.13
*Noncardiac anomaly*	30 (11%)	3 (9%)	0.73

Table 2. Echocardiographic Data

Variables	No ECMO (n=263)	ECMO (n=32)	p-value
*Tricuspid valve z-score*	-2.5 ± 2.0	-3.1 ± 2.3	0.15
*Tricuspid valve regurgitation* *None* *Mild* *Mild to moderate* *Moderate* *Moderate to severe* *Severe* *Not available*	97 (37%) 57 (22%) 27 (10%) 37 (14%) 8 (3%) 9 (3%) 28 (11%)	14 (44%) 4 (13%) 3 (9%) 2 (6%) 1 (3%) 1 (3%) 7 (22%)	0.43
*Right ventricular morphology* *Unipartite* *Bipartite* *Tripartite* *Not available*	30 (11%) 28 (11%) 80 (30%) 125 (48%)	9 (28%) 3 (9%) 8 (25%) 12 (38%)	0.07
*Right ventricular size* *Mild hypoplasia* *Moderate hypoplasia* *Severe hypoplasia* *Normal* *Mild dilation* *Moderate dilation* *Severe dilation*	33 (13%) 69 (26%) 136 (52%) 13 (5%) 2 (1%) 4 (2%) 2 (1%)	2 (6%) 4 (13%) 22 (69%) 4 (12%) 0 (0%) 0 (0%) 0 (0%)	0.21
*Estimated right ventricular pressure*	100.7 ± 32.6	91.0 ± 23.9	0.44
*Pulmonary atresia type* *Membranous* *Muscular* *Not available*	139 (53%) 31 (12%) 93 (35%)	11 (34%) 5 (16%) 16 (50%)	0.14
*Pulmonary valve annulus z-score*	-2.3 ± 1.4	-1.9 ± 1.5	0.45
*Right pulmonary artery z-score*	-1.0 ± 0.9	-0.5 ± 0.7	0.01
*Left pulmonary artery z-score*	-0.8 ± 1.0	-0.2 ± 1.0	0.01
*Left ventricular ejection fraction*	64.4 ± 10.0	65.4 ± 6.8	0.67
*Presence of right coronary artery*	223 (85%)	23 (72%)	0.06
*Presence of left coronary artery*	225 (86%)	26 (82%)	0.51
*Evidence of to-fro flow in the coronary circulation*	145 (55%)	20 (62%)	0.52
*Presence of ventriculo-coronary connections*	98 (46%)	15 (60%)	0.17

Table 3. Catheterization Data

Variables	No ECMO (n=263)	ECMO (n=32)	p-value
*Presence of ventriculocoronary connections by catheterization*	111 (51.6%)	19 (76%)	0.02
*Coronary artery stenosis by catheterization*	37 (14%)	11 (34%)	0.56
*Coronary ostial atresia by catheterization*	24 (11%)	7 (28%)	0.01

Table 4. Regression Results Modeling Risk for Extracorporeal Membrane Oxygenation

Variables	Beta-coefficient	Odds ratio	p-value
*Left pulmonary artery z-score*	0.72	2.0	< 0.01
*Ventriculocoronary connections by catheterization*	1.25	3.4	0.03

Multivariable analyses, use of ECMO

Multivariable regression analyses demonstrated that higher left pulmonary artery z-score by echocardiography (beta-coefficient 0.72) and the presence of ventriculocoronary connections by cardiac catheterization (beta coefficient 1.25) were significantly associated with increased use of ECMO (Table 4).

The resulting model was (left pulmonary artery z-score *0.82 + presence of ventriculocoronary connections *1.25). The resulting risk score had an area under the curve of 0.71 (p < 0.01) for prediction of ECMO use. The optimal cutoff for the score was 0.30 which had a sensitivity of 0.81 and specificity of 0.58. A score of less than − 2.0 was associated with a 0% risk of extracorporeal membrane oxygenation. A score of greater than or equal to -2.0 but less than 0 was associated with a 4% risk. A score of greater than or equal to 0 but less than 2.0 was associated with a 16% risk. And a score of greater than 2.0 was associated with a 25% risk.

Univariable analyses, inpatient mortality in those receiving extracorporeal membrane oxygenation

Among the 32 patients with PAIVS who received ECMO, 15 (46%) experienced inpatient mortality. Those who experienced inpatient mortality tended to have a higher gestational age (Table 5) and a lower pulmonary valve z-score (Table 6). No significant differences were noted in cardiac catheterization data (Table 7).

Table 5. Characteristics of Patients Requiring Extracorporeal Membrane Oxygenation

Variables	No mortality (n=17)	Mortality (n=15)	p-value
*Prenatal diagnosis*	16 (94%)	11 (73%)	0.10
*Gestational age (weeks)*	36.9 ± 2.7	38.5 ± 0.9	0.04
*Weight (kg)*	2.7 ± 0.6	2.9 ± 0.3	0.16
*Height (cm)*	46.9 ± 4.5	48.4 ± 2.7	0.29
*Gender* *Male* *Female*	7 (41%) 10 (59%)	6 (40%) 9 (60%)	0.94
*Ethnicity* *Hispanic* *Not Hispanic*	5 (29%) 12 (71%)	6 (40%) 9 (60%)	0.52
*Chromosomal abnormality*	1 (6%)	0 (0%)	0.34
*Noncardiac anomaly*	2 (12%)	1 (7%)	0.62

Table 6. Echocardiographic Characteristics in Patients Requiring Extracorporeal Membrane Oxygenation

Variables	No mortality (n=17)	Mortality (n=15)	p-value
*Tricuspid valve z-score*	-2.6 ± 2.3	-3.5 ± 2.4	0.34
*Tricuspid valve regurgitation* *None* *Mild* *Mild to moderate* *Moderate* *Moderate to severe* *Severe* *Not available*	6 (35%) 2 (12%) 2 (12%) 0 (0%) 1 (6%) 1 (6%) 5 (29%)	8 (54%) 2 (13%) 1 (7%) 2 (13%) 0 (0%) 0 (0%) 2 (13%)	0.44
*Right ventricular morphology* *Unipartite* *Bipartite* *Tripartite* *Not available*	3 (18%) 3 (18%) 5 (29%) 6 (35%)	6 (40%) 0 (0%) 3 (20%) 6 (40%)	0.22
*Right ventricular size* *Mild hypoplasia* *Moderate hypoplasia* *Severe hypoplasia* *Normal* *Mild dilation* *Moderate dilation* *Severe dilation*	2 (12%) 2 (12%) 11 (64%) 2 (12%)	0 (0%) 2 (13%) 11 (74%) 2 (13%)	0.59
*Estimated right ventricular pressure*	97.6 ± 7.3	86.0 ± 32.2	0.57
*Pulmonary atresia type* *Membranous* *Muscular* *Not available*	6 (35%) 4 (24%) 7 (41%)	5 (33%) 1 (7%) 9 (60%)	0.36
*Pulmonary valve annulus z-score*	-0.9 ± 1.3	-3.0 ± 0.8	< 0.01
*Right pulmonary artery z-score*	-0.5 ± 0.4	-0.5 ± 0.9	0.85
*Left pulmonary artery z-score*	-0.1 ± 0.9	-0.3	0.75
*Left ventricular ejection fraction*	64.7 ± 9.0	66.2 ± 3.7	0.66
*Presence of right coronary artery*	11 (65%)	12 (80%)	0.30
*Presence of left coronary artery*	13 (77%)	13 (87%)	0.42
*Evidence of to-fro flow in the coronary circulation*	11 (65%)	9 (60%)	0.78
*Presence of ventriculo-coronary connections*	7 (58.3)	8 (61.5%)	0.87

Table 7. Catheterization Characteristics in Patients Requiring Extracorporeal Membrane Oxygenation

Variables	No mortality (n=17)	Mortality (n=15)	p-value
*Presence of ventriculo-coronary connections by catheterization*	9 (75%)	10 (77%)	0.19
*Coronary artery stenosis by catheterization*	2 (17%)	5 (39%)	0.22
*Coronary ostial atresia by catheterization*	3 (25%)	4 (31%)	0.74

Multivariable analyses, inpatient mortality in those receiving ECMO

Multivariable regression analyses did not demonstrate significant associations of any independent variables with inpatient mortality.

This multicenter study of contemporary outcomes in PAIVS demonstrates that 11% of children with were supported with ECMO. Patient anatomic factors including higher left pulmonary artery z-score and the presence of ventriculocoronary connections identified by imaging in the first week of life were associated with increased likelihood of receiving ECMO, with an area under the curve of 0.71. Of those receiving ECMO, 54% survived to discharge, which is comparable to the overall pediatric cardiac ECMO mortality[4]. Multivariable analyses did not identify any factors associated with inpatient mortality, but this analysis was limited by a small sample size.

PAIVS is a lesion in which children initially have parallel circulation (wherein the saturation of blood going to the pulmonary and systemic circulations is equal). Parallel circulation is associated with increased risk of morbidity and mortality due to the inherent systemic venous desaturation and the coupling of the systemic arterial saturation to both pulmonary venous and systemic venous saturation along with the total cardiac output having to be distributed amongst the pulmonary and systemic circulations[3].

Abnormalities of the coronary arterial circulation and subsequent limited myocardial oxygen delivery along with parallel circulation, can lead to inadequate systemic and myocardial oxygen delivery[5]. This inadequacy of oxygen delivery may warrant ECMO. Greater branch pulmonary artery z-score and presence of ventriculocoronary connections were found to be associated with increased risk for extracorporeal membrane oxygenation. Larger pulmonary arteries may lead to high pulmonary blood flow, specifically when parallel circulation is present and the distribution of cardiac output between the pulmonary and systemic circulations depends on the relative resistances of the two circuits. Greater pulmonary blood flow leads to less systemic blood flow which may lead to a lower organ oxygen delivery. It is unclear why the left pulmonary artery size is specifically associated with ECMO utilization whenthe right pulmonary artery size is not.

Ventriculocoronary connections may affect ECMO outcome as ventriculocoronary connections are typically present in conjunction with coronary artery stenosis or coronary ostial atresia. In addition, there could be a specific risk of decompressing the RV which may impede coronary blood flow. These may limit myocardial oxygen delivery and lead to impaired ventricular function and subsequent limitations in systemic oxygen delivery[6]. Additionally, impaired myocardial oxygen delivery may increase the risk of arrhythmia. All these factors may impact not only hospital mortality but also long-term outcomes in these patients[7]. Of note the presence of ventriculocoronary connections by catheterization but not coronary abnormalities on echocardiography was associated with increased risk of use of ECMO in this population. Thus, the clinical implication of findings between the two modalities does differ.

Once receiving ECMO, children with PAIVS are anecdotally thought to have uniformly poor outcome and high mortality. Our study contrasts this notion and quantifies survival in these patients, finding that over half of these patients survive to hospital discharge. Therefore, while pursued with caution, ECMO should not be considered contraindicated in patients with PAIVS [4].

Identifying patients at increased risk of deterioration and use of ECMO may help improve vigilance for these patients and help guide decision making and timely initiation of therapies and parental counseling. Ideally, anticipation and prevention would mitigate some associated morbidity and mortality.

These analyses offer novel insight in the form of accurate assessment for risk of ECMO use in children with PAIVS. They also offer novel insight into risk of mortality for these patients once on ECMO. However, the findings must be evaluated in the context of study limitations. These analyses take into account early patient characteristics, anatomic findings from the initial echocardiogram, and initial cardiac catheterization data but not data regarding specific medical, surgical, or catheter-based interventions. This was done purposefully to determine the utility of a system that would allow for early risk assessment. Surprisingly, this approach yielded good accuracy. There is always a possibility of variability in echocardiographic measurements such as those utilized in this study. This is a limitation, as there was no central core echocardiography center. Likewise, assessment of many echocardiographic variables is subjective. We limited the use of such subjective assessments although variables such as tricuspid regurgitation, right ventricular morphology, and right ventricular size remain subjective in nature. Sample size and missing values limit the analyses as well. Despite this study being a multicenter study with a relatively large sample size for this relatively infrequent congenital malformation, there are inherent limitations related to its retrospective nature. While sample size was adequate for assessing variables associated with need for ECMO, the sample size of patients using ECMO was much smaller limiting ability to find variables associated with inpatient mortality once using ECMO.

In a multicenter contemporary cohort of patients with PAIVS, 11% received ECMO. Risk factors for ECMO use included higher left pulmonary artery z-score on initial echocardiography and the angiographic presence of ventriculocoronary connections. Survival of patients with PAIVS using ECMO is better than historically perceived, as 54% survived to discharge. These findings provide important data in a contemporary cohort of PAIVS patients and help clinicians with risk assessment and counseling.

Acknowledgements

None

Conflicts of Interest

All authors have disclosed that they do not have any conflicts of interest

Sources of Funding

No sources of Funding to report

Lawley C, Hockey K, Yeo LL, Liava'a M, Roberts P (2022) Increasing Use of Neonatal Catheter Intervention for Pulmonary Atresia With Intact Ventricular Septum: Management Trends From a Single Centre. Heart, lung & circulation 31: 549-558
Walsh MA, Asoh K, Van Arsdell GS, Humpl T (2010) Critical care outcomes in pulmonary atresia and intact ventricular septum undergoing single-ventricle palliation. Cardiology in the young 20: 290-296
Petrucci O, O'Brien SM, Jacobs ML, Jacobs JP, Manning PB, Eghtesady P (2011) Risk factors for mortality and morbidity after the neonatal Blalock-Taussig shunt procedure. The Annals of thoracic surgery 92: 642-651; discussion 651-642
Allen KY, Allan CK, Su L, McBride ME (2018) Extracorporeal membrane oxygenation in congenital heart disease. Semin Perinatol 42: 104-110
Freedom RM, Anderson RH, Perrin D (2005) The significance of ventriculo-coronary arterial connections in the setting of pulmonary atresia with an intact ventricular septum. Cardiology in the young 15: 447-468
Ashburn DA, Blackstone EH, Wells WJ, Jonas RA, Pigula FA, Manning PB, Lofland GK, Williams WG, McCrindle BW (2004) Determinants of mortality and type of repair in neonates with pulmonary atresia and intact ventricular septum. The Journal of thoracic and cardiovascular surgery 127: 1000-1007; discussion 1007-1008
Muneuchi J, Watanabe M, Sugitani Y, Doi H, Furuta T, Kobayashi M, Ezaki H, Ochiai Y (2022) Long-Term Outcomes After an Individualized Strategy in Patients with Pulmonary Atresia and Intact Ventricular Septum. Pediatric cardiology 43: 435-442

Supplemental Table 1 is not available with this version

No competing interests reported.

Extracorporeal Membrane Oxygenation in Children with Pulmonary Atresia and Intact Ventricular Septum: Mortality and Associated Outcomes

Status:

Version 1

Abstract

Introduction

Methods

Results

Conclusion

Introduction

Methods

Patient identification

Variables of interest

Statistical analyses

Results

Univariable analyses, use of ECMO

Multivariable analyses, use of ECMO

Univariable analyses, inpatient mortality in those receiving extracorporeal membrane oxygenation

Multivariable analyses, inpatient mortality in those receiving ECMO

Discussion

Conclusion

Declarations

References

Supplemental Table 1

Additional Declarations

Status:

Version 1