Timing of Balloon Atrial Septostomy in Patients with d-TGA and Association with Birth Location and Patient Outcomes

Patients with d-looped transposition of the great arteries (d-TGA), especially those without an adequate atrial septal defect, can experience severe hypoxemia and hemodynamic compromise in the neonatal period. This can be mitigated by urgent balloon atrial septostomy (BAS). However, some patients with d-TGA are born at centers without this capability. The aim of this retrospective study of d-TGA patients who had urgent or emergent BAS at our institution between 2010 and 2021 was to evaluate time from birth to BAS for infants born at a tertiary care center as compared to those requiring transport from other institutions and to examine correlation between time to BAS and patient outcomes. Our primary outcome was time from birth to BAS. Secondary outcomes included hospital and ICU length of stay, mortality, and evidence of pulmonary or neurologic abnormalities including pulmonary hypertension, abnormal neuroimaging, or seizures. Of 96 patients, 67 (70%) were born at our institution. The median time to BAS was 4 h for patients born at our institution vs. 14.1 h for those born elsewhere (p < .0001). A longer time from birth to BAS was associated with longer ICU (r = 0.21, p = 0.046) and hospital length of stay (r = 0.24, p = 0.02) and increased likelihood of elevated right ventricular pressure on post-operative discharge echocardiogram (p = 0.01). There were no differences in mortality between the groups. Therefore, prenatal planning for patients with known d-TGA should include a delivery plan with access to urgent BAS.


Introduction
Transposition of the great arteries is the most common form of cyanotic congenital heart disease with d-looped transposition (d-TGA) accounting for 5-7% of all congenital heart defects [1]. In the initial postnatal period, patients with d-TGA may be subject to profound hypoxemia and resulting hemodynamic compromise if intracardiac mixing of blood is inadequate. Given the high-risk postnatal period, studies have shown that expedited access to medical and interventional therapies has improved outcomes and survival for neonates with d-TGA [2]. While pre-operative management and surgical timing vary by institution, pulse oximetry monitoring, transthoracic echocardiography, and initiation of prostaglandin E1 (PGE) are generally recognized as important components of initial management. Additionally, for infants with prohibitive hypoxemia and a restrictive atrial septum, balloon atrial septostomy (BAS) is a well-established method of improving intracardiac mixing and hemodynamic status [3]. However, not all infants with d-TGA and inadequate atrial mixing are born at centers with the ability to rapidly perform BAS procedures. Prior studies have attempted to look at outcomes related to birth location, but these data have been conflicted and often limited to broad outcomes like mortality [4][5][6].
This study aimed to evaluate our center's experience with timing of BAS in infants with d-TGA comparing infants born at our center and those requiring transport from other institutions and how this timing and birth location relates to overall outcomes. In addition, besides mortality, we examined other outcomes including hospital and intensive care unit (ICU) length of stay, as well as markers of pulmonary and neurologic sequelae of prolonged hypoxemia.

Study Design
After obtaining Institutional Review Board approval, we performed a single-center retrospective study of all neonates with d-TGA or d-TGA physiology who underwent urgent or emergent BAS at our institution between 2010 and 2021. An urgent BAS was defined as a BAS occurring within the first 48 h of life (or upon arrival to our center if more than 48 h of life) with evidence of poor oxygenation (SpO 2 < 70% on room air, FiO 2 requirement > 40% with or without mechanical ventilation) but without hemodynamic instability. An emergent BAS was defined as a BAS occurring within the first 48 h of life (or upon arrival to our center) with evidence of significantly compromised oxygenation (PaO 2 < 30 mmHg or SpO 2 < 70% on room air with mechanical ventilation and an FiO 2 requirement > 60%) and/or hemodynamic instability (pH < 7.3 or requiring ionotropic support). Patients were excluded from the study if they underwent an elective BAS at more than 48 h of life, did not undergo a BAS, or had a BAS at an outside institution prior to transfer to our center. All patients had their diagnosis confirmed via postnatal echocardiography. Patients were identified using our internal cardiac catheterization database and the institutional Electronic Medical Record Search Engine (EMERSE) [7].

Variables
In addition to patient demographics, gestational age, and birth characteristics, prenatal factors were collected including location of prenatal care and for those patients with a prenatal diagnosis, location of fetal echocardiogram, fetal diagnosis and atrial septal characteristics, and location of fetal cardiology consultation. Postnatal but pre-BAS variables included need for intubation, ionotropic support, or cardiopulmonary resuscitation (CPR) as well as highest FiO 2 , lowest and highest oxygen saturation on pulse oximetry, and laboratory values including pH, PaO 2 , and lactate. Details of the postnatal echocardiogram were also collected including the specific d-TGA physiology, atrial septal characteristics, presence of a patent ductus arteriosus (PDA), and ventricular function. BAS procedural characteristics were collected including the time from birth to ASD creation and procedural complications. Duration of mechanical ventilation, ICU and hospital length of stay, need for extracorporeal membrane oxygenation (ECMO), and survival were collected for both the time period between BAS and definitive surgical repair and time period after definitive surgical repair. We also collected the need for inhaled nitric oxide (iNO) during these time periods as well as right ventricular (RV) pressure on discharge echocardiogram. Markers of neurologic injury as indicated by clinical or subclinical seizures or abnormal neuroimaging were also collected. Lastly, long-term outcomes including vital status at most recent follow-up were collected for all patients.

Outcomes
The primary outcome of the study was time from birth to BAS. Secondary outcomes included mortality, hospital and ICU length of stay, duration of mechanical ventilation, evidence of pulmonary hypertension (as defined by pulmonary vasodilator requirement, either inhaled nitric oxide or oral therapy, or elevated RV pressures on postoperative discharge echocardiogram), and neurologic abnormalities including abnormal neuroimaging or seizures.

Statistical Analysis
Data are described as frequencies with percentage (%) for categorical variables and median with interquartile range (IQR) or mean ± standard deviation for continuous variables. Patient demographics, birth and clinical characteristics, and outcomes were compared between infants born at our center and those delivered elsewhere using Chi-square test or Fisher's exact test for categorical variables and Wilcoxon rank sum test or two-sample t test for continuous variables. Univariate associations of time from birth to BAS with birth location and patient outcomes were examined using Wilcoxon rank sum test for categorical variables and Spearman correlation coefficient, r, for continuous variables. All analyses were performed using SAS Version 9.4 (SAS Institute Inc., Cary, NC), with statistical significance set at a P value < 0.05 using a two-sided test.

Results
Of 217 patients with d-TGA or d-TGA physiology who underwent neonatal surgical intervention at our institution during the study period, a total of 96 (44%) patients required an emergent or urgent BAS and were included in the analysis. The patient demographics are shown in Table 1. The majority of patients were prenatally diagnosed with d-TGA, and those infants are described in Table 2. Of the prenatally diagnosed infants, 97% received a prenatal consultation at our congenital heart center. Only two of the patients who had prenatal consultation at our institution (U of M) delivered outside our institution, both due to maternal presentation for urgent induction of labor. One infant was diagnosed prenatally but was not referred for fetal consultation the mother was not counseled to deliver at U of M. Of the 29 patients without a prenatal diagnosis, the majority (n = 23, 79%) received adequate prenatal care. Only 1 patient had limited prenatal care due to late presentation. In the other 5, prenatal care was unknown/not documented. The infants born at other hospitals had a median travel distance of 99 miles (IQR 56-115) and the transport time was a median of 10.3 h (IQR 6.8-12.0). When comparing the infants born outside our institution with (n = 3) and without (n = 26) a prenatal diagnosis, the transport time was a median of 6.4 h (IQR 4.0-6.8) vs. 10.7 h (IQR 7.9-13.5), respectively, suggesting slightly faster transport time for the prenatally diagnosed infants, though there were no specific plans made prenatally for expedited transfer of infants born at outside institutions.
The clinical characteristics prior to BAS are described in Table 3. Infants delivered outside our institution had lower minimum PaO 2 (p = 0.02) and were more likely to require ionotropic support (p < 0.0001). The median APGAR score at both 1 min and 5 min was 8, and there was no difference in the APGAR scores of infants born at U of M compared to those delivered elsewhere. There were no significant differences in the postnatal echocardiogram findings between the groups. Figure 1 demonstrates the time from birth to BAS for the two groups. Infants who were delivered outside our institution had the median time from birth to BAS that is nearly 10 h longer than those born at U of M. The median time of arrival in the lab was 14 h in patients born elsewhere vs. 3.4 h in patients born at U of M but the time from procedure start to septostomy and total procedure duration are  comparable between the two groups, as shown in Table 4.
Overall, the median time from procedure start to septostomy was 16 min and the median total procedure duration was 52 min. Additionally, the number of procedures performed during typical daytime operations versus after hours was similar between groups. Septostomy procedural complications were rare, occurring in 5 patients (5%). While it was not significantly different between the groups, the two patients who had cardiac arrests during their septostomy procedures were both delivered outside our institution. The other major complication was a pericardial effusion requiring treatment (n = 1). Minor complications included transient 2:1 heart block, transient supraventricular tachycardia (SVT), and prolonged bleeding from the femoral venous access site (n = 1 each).
Outcomes of interest were evaluated both after septostomy/prior to surgery as well as after surgery and prior to discharge. There were no significant differences in the post-septostomy/pre-surgery hospital courses based on time to septostomy ( Table 5) or location of birth. Of note, the median age at time of surgery and median time between septostomy and surgery were the similar between groups (5 and 4 days for patients born at U of M, 5 and 5 days for patients born elsewhere, p = 0.27 and p = 0.89, respectively).
There were four infants who did not undergo surgical repair during their initial admission. All four of these patients had d-TGA/VSD with pulmonary stenosis. Two had a balanced circulation, they were discharged home and underwent Rastelli at 3 and 5 months of age. One patient had a PDA stent prior to discharge, and one was noted to have crossing atrioventricular valvar chordae so ultimately underwent a superior cavopulmonary anastomosis (hemi-Fontan at our center). For the infants who underwent definitive cardiac surgery, a longer time from birth to BAS was associated with a longer ICU length of stay (r = 0.21, p = 0.046) as well as a longer total hospital length of stay (r = 0.24, p = 0.02) ( Table 6). There was no difference in the use of inhaled or oral pulmonary vasodilators; however, infants with a longer time from birth  to BAS were more likely to have elevated RV pressure on discharge echocardiogram (p = 0.004). There were no differences in frequency of seizures or abnormal neuroimaging based on time to BAS or birth location. During median 3.8 years of follow-up (IQR 1.3-6.8), there were 6 (6%) deaths and the majority of them occurred within the first year of life, with 5 of the 6 deaths occurring within 6 months of BAS. There was no significant difference in mortality with longer time to BAS or between the group of patients born at U of M and those delivered outside our institution.

Discussion
In this study, we found that newborns with d-TGA who are delivered outside of a tertiary care center with the ability to perform a BAS had a significantly longer time from birth to BAS. In addition to these patients having lower PaO 2 and receiving increased inotropic support prior to BAS, this increased time from birth to BAS was associated with a longer ICU and hospital length of stay as well as an increased likelihood of elevated RV pressure on discharge echocardiogram. This study adds to literature related to birth location and d-TGA outcomes by evaluating the role of timing of BAS in the overall patient outcomes. Previous studies in the United States evaluating birth location and/or prenatal diagnosis in this population have focused on mortality as the primary outcome measure [8]. However, in the current surgical era, mortality after arterial switch operation is rare. In addition, most of those studies did not evaluate outcomes based on the timing of BAS. Some prior European and Australian studies have shown no difference in outcomes related to time to BAS, but other data suggest that there is an increased rate of adverse events, higher rates of mechanical ventilation, ECMO utilization, and significant hypoxemia in infants transported from referring facilities [4-6, 9, 10]. In our study, the degree of hypoxia was worse in patients transported from outside facilities. And while duration of mechanical ventilation was not found to be impacted by location of birth or time to BAS in our study, the increased frequency of elevated RV pressure on discharge echocardiogram in those infants who had a later BAS could indicate lasting effects of prolonged, severe neonatal hypoxemia on the pulmonary vasculature. We did not identify any acute neurologic sequalae related to timing of BAS. Of note, more patients who had abnormal head imaging underwent earlier BAS. It is possible that those who undergo very rapid BAS (within the first few hours of life), may represent a group of patients with more severe hypoxia or hemodynamic instability. It is not possible to decipher if severity of disease or timing of BAS is independent determinants of abnormal neuroimaging from this study. One limitation to our study is a lack of a standard approach to neuroimaging for the d-TGA patient population. Many patients (44% after BAS and 71% after surgical repair) had no neuroimaging, and those that did were often limited to a cranial ultrasound which cannot detect subtle changes related to hypoxia. Other studies suggest that newborns with a prenatal diagnosis of critical congenital heart disease (including d-TGA) have less pre-operative brain injury on magnetic resonance imaging (MRI) and more robust brain development than those with postnatal diagnosis [11]. Depending on the type of congenital heart disease, postnatally diagnosed infants may have longer duration of hypoxemia, and this may have a similar effect for infants with d-TGA with a longer time from birth to BAS. Therefore, longer-term assessment of neurodevelopment and neurocognitive function in these patients, or more global utilization of MRI to assess for brain injury, may help to identify important neurologic outcomes. The role of prenatal diagnosis using fetal echocardiography is crucial for these patients. While d-TGA has been historically challenging to diagnose prenatally, recent studies cite a detection rate of 50-80%, similar to what was seen in our study at 70% [12,13]. Prenatal detection allows for counseling of expectant mothers and often, leads to the birth of these infants at tertiary care centers with the ability to perform urgent interventions such as a BAS. In our study, 72% of patients were prenatally diagnosed and 99% of the infants with a prenatal diagnosis delivered at our institution where access to BAS is readily available. It is the practice at our institution to counsel all expectant mothers to deliver at our center regardless the appearance of the atrial septum in utero or presence of a VSD. Improved fetal detection and delivery planning has been shown to improve overall outcomes for these infants [13]. Some of this improvement in outcomes may be directly related to improved access to and shortened time from birth to BAS.
Approximately 44% of all patients with d-TGA physiology treated at our center during the study period underwent emergent or urgent BAS. This is consistent with what was seen in a recent multicenter analysis from O'Byrne et al. which found an average frequency of BAS of 49% [14]. Their study noted wide practice variation in the use of BAS amongst various hospitals with some performing them on only 20% of patients and others on more than 80% of patients. They also described that BAS was associated with shorter total length of stay and lower hospital cost, which is consistent with our findings that earlier BAS was associated with shorter duration of ICU and hospital length of stay. Of note, another recent study evaluating a large group of neonates with d-TGA reported that 28% of neonates with d-TGA and VSD required a septostomy, in our study, 25% of the infants requiring a septostomy had a VSD [15]. In that study, 33% of infants not requiring a BAS had a VSD and this was a statistically significant difference [15]. This suggests that while need for a BAS may be less common in the patients with a VSD, there is still a sub-population of them that do require a septostomy.
In addition to the limitations related to evaluating neurologic sequelae in our study, our sample size and strong correlation between prenatal diagnosis and birth at our institution makes it difficult to distinguish the impact of those variables completely independently. In addition, as a retrospective study, it is not possible to understand or account for subtle differences in patient care management which could have impacted timing of BAS. While a multicenter study may help find more associations between timing of BAS and patient outcomes, a prospective study would likely be necessary to delineate the impact of birth location and timing of BAS as independent factors.

Conclusions
While some infants may continue to be diagnosed with d-TGA postnatally and, thus, require transport to tertiary care centers, infants prenatally diagnosed with d-TGA benefit from delivery planning that includes birth at a tertiary care center with access to interventional therapies including the ability to perform a rapid BAS. Further research to evaluate neurodevelopmental outcomes in this patient population has the potential to uncover additional effects of prolonged postnatal hypoxia.
Author Contributions CT: pediatric cardiology fellow, data collection, abstract and manuscript preparation. SY: statistician, data analysis. RL: data manager, creation, and management of database. JDZ: faculty mentor, abstract and manuscript preparation and editing.
Funding No funding was received to assist with the preparation of this manuscript.

Data Availability Not applicable.
Code Availability Not applicable.

Consent to Participate Not applicable.
Consent for Publication Consent for publication of this manuscript is given.
Ethical Approval Institutional review board.