D-Transposition of the great arteries with restrictive foramen ovale in the fetus: the dilemma of predicting the need for postnatal urgent balloon atrial septostomy

The simple dextro-Transposition of the Great Arteries with intact ventricular septum (d-TGA) is one of the most common cyanotic heart defects with a prevalence of 2–3 of 10.000 live births. It accounts for 5–7% of all congenital heart defects [1, 2]. Simple d-TGA is characterized by atrioventricular concordance and ventriculo-arterial discordance, leading to redistribution of blood flow with subsequent development of severe hypoxemia and rapid hemodynamic compromise postnatally. Both, fetal pulmonary circulation and ductus arteriosus become increasingly sensitive to the high oxygen content toward the end of gestation, which may predispose to a restriction of the foramen ovale (FO) and ductus arteriosus at term by increase of left atrial pressure leading to restrictive FO flow [3‐7]. In addition, the increased right ventricular outflow and decreased antegrade flow through the arterial duct lead to a reduced right-to-left shunt across the FO and to a stagnation of FO growth during the 3rd trimester [8, 9]. Consequently, the FO may be significantly restrictive in a significant proportion of newborns, making an urgent balloon atrial septostomy (BAS) immediately after birth inevitable in about 28% of newborns [9, 10]. Indisputably, prenatal diagnosis of critical d-TGA with restrictive FO with planned delivery in a tertiary center with cardiac catheter stand-by, allowing immediate BAS after birth, improves neonatal survival, whereas a significant delay of diagnosis and intervention postnatally may result in severe life-threatening neonatal hypoxia [11‐13].

Several conventional prenatal echocardiographic parameters were previously evaluated with contradictory sensitivity and specificity in predicting restrictive FO and the likelihood for urgent BAS [6, 7, 9, 14‐21]. Recently, modern techniques on cardiac functional assessment like spectral pulsed wave tissue-Doppler imaging, speckle-tracking echocardiography, or fetal cardiac magnetic resonance imaging have demonstrated a potentially higher sensitivity for detection of subtle subclinical changes in myocardial dysfunction [7, 22‐24], but data on functional assessment in fetuses with d-TGA are limited to only one single small cohort of only 13 fetuses without validation in larger prospective trials [7]. Additionally, the calculation of functional cardiac indices is time-consuming and requires highly specified instrumental equipment. Therefore, the use of these technologies is so far limited to a very few specialized centers and research purposes only.

This study aimed to describe our experience with the prenatal prediction of normal or restrictive FO anatomy in fetuses with d-TGA. We first reviewed the reliability of our own former prenatal ultrasound diagnoses made at the time of the patients’ visits in our centers prior to birth. Second, we re-assessed all stored echocardiographic videos and still images and correlated them with the outcome. Thereby, we retrospectively identified the most reliable echocardiographic parameters for restrictive FO. Finally, we described the predictive value of these parameters by calculating their sensitivity and specificity.

During the study period, 87 fetuses with isolated d-TGA with intact ventricular septum and known outcome were diagnosed and delivered in one of our two participating centers. 42 cases had to be excluded from the study. Reasons for exclusion were either preterm delivery before the completed 34th week or lack of suitable prenatal echocardiographic videos or still images for reliable re-analysis. Finally, 45 cases were included in this series. Fetal or neonatal genetic testing, including karyotype and 22q11 microdeletion, was normal in all cases. There was no termination of pregnancy and no intrauterine or neonatal death in our cohort.

Cesarean section was performed in 26 (57.7%) of 45 cases. Reasons for surgical delivery were either prenatal prediction of restrictive FO in 15 cases, obstetrical reasons in seven cases and maternal request in four cases. Median gestational age at delivery was 39 weeks (range, 36–41 weeks).

Among the 45 fetuses, 22 neonates had restricted FO at birth and underwent urgent BAS within the first 24 h of life, whereas 23 neonates had normal FO size at birth. Among the latter, four neonates unexpectedly showed inadequate interatrial mixing despite a normal FO anatomy. They did not achieve sufficient oxygen saturation, rapidly developed hypoxia, and also required urgent BAS (‘bad mixer’). Finally, 26 (57.8%) neonates required urgent BAS within the first 24 h of life (urgent BAS group), whereas 19 (42.2%) neonates achieved good oxygen saturation and did not underwent urgent BAS (non-urgent BAS group) (Fig. 5).

Among the 4 “bad mixers”, all had normal FO appearance with normal FO diameters between 8 and 9 mm, all had a normal FO flap mobility of 30–50° and none of them had a restrictive DA prenatally. Additionally, in 3 of those 4 cases, the pulmonary veins were also assessed prenatally and also showed a normal, plateau-like flow pattern with peak velocities between 20 and 30 cm/s (Table 3, patients no 23–26). All those prenatal findings were confirmed on first postnatal echocardiography that was performed immediately after birth.

Indisputably, prenatal prediction of a restrictive FO with planned delivery in a tertiary cardiac center and consecutively cardiac intervention within 24 h after birth improves survival, whereas a significant delay of diagnosis and intervention may result in severe and life-threatening neonatal hypoxia. Several prenatal cardiac parameters and classification systems have previously been evaluated to predict premature restriction of the FO and the likelihood of urgent BAS [6, 7, 9, 14‐21]. Table 5 shows an overview of the previously published series and the included cardiac parameters.

These series showed contradictory reliabilities of the different cardiac parameters in predicting the postnatal FO anatomy and the likelihood of urgent BAS. These difficulties in predicting postnatal hypoxia may be explained by several factors. First, fetal hemodynamic parameters in late pregnancy are influenced by several factors other than those related to FO diameter, such as breathing movements, greater sensitivity to oxygen, and physiological right ventricular preponderance [15]. In addition, all published data relied on retrospective studies and included different cardiac parameters that were obtained at different gestational ages with highly variable time intervals between last prenatal echocardiography and birth. Moreover, postnatal oxygen saturation cut-off values differed, and the indication to perform an urgent BAS was not standardized in previously published reports, making those data only to a limited extend comparable.

At first, Maeno et al. described 16 fetuses with isolated d-TGA in 1999 [14]. They assessed the appearance of the FO and the FO flap as well as the DA diameter, but last sonographic assessment prior to delivery was performed at a mean gestational age of 31 weeks without performing an additional late echocardiography. They concluded that the sensitivity to predict a restrictive FO is low. But in a significant subset of fetuses, FO restriction develops not until the late third trimester and last echocardiography at 31 weeks of gestation may be too early to detect significant restriction of the FO in these cases. As the critical period is near term, we strongly recommend to perform an additional late echocardiography after 35 weeks of gestation in all fetuses with d-TGA to once again re-assess the FO anatomy and to re-evaluate the likelihood of postnatal urgent BAS.

Since then, most recent series including a large meta-analysis by Buca et al. described a significantly smaller FO diameter in fetuses that required urgent BAS [6, 9, 15, 18, 20, 21], but a FO diameter cut-off value to reliably predict a restrictive FO was only described in one single series [20]. In this series, 60 fetuses with isolated d-TGA were retrospectively assessed and the authors concluded that a FO diameter of more than 6.5 mm predicts a normal FO with 100% sensitivity and a false-positive rate of 45%. We could confirm, that a FO diameter of less than 7 mm is highly predictive for a restrictive FO and need for urgent BAS (p < 0.01).

Inconsistent findings were published with regard to the septum primum appearance and the FO flap movement. Some authors stated that a redundant, ‘spinnaker-like’ [9, 18, 20, 21] or flat [6, 19] FO flap significantly predicts a restrictive FO, whereas other authors could not confirm this finding [14‐16]. Vigneswaran et al. postulated that a redundant FO may be explained as a variant of the normal third trimester atrial septum deviation rather than as a pathological marker [6]. In contrast, a redundant FO flap may cause an occlusion of the FO orifice in some cases as it may mimic primary restriction of FO in cases of an aneurysmal septum primum. Some authors described a hypermobile FO flap as a significant predictor for restrictive FO [16, 21], whereas others could not support this finding and assumed that this may be related to simulated breathing movements or hiccoughs rather than to a restrictive FO [6]. In our series, both a fixed and hypermobile flap mobility were the most significant predictors for a restrictive FO and the need for an urgent BAS (p = 0.035 and p = 0.014, respectively).

Only one previously published study assessed the flow velocity in the pulmonary veins (PV) and described a significantly increased maximum systolic flow in the PV in restrictive FO. They stated, that a cut-off value of 41 cm/s maximum flow velocity provided maximum specificity of 100% with a positive predictive value of 100% [19]. In accordance, we could confirm that the maximum flow velocity in the PV is increased in restricted FO, but this difference reached no statistical significance (p = 0.216) in our data and no cut-off value to reliably predict the restrictive FO could be identified. In addition, the flow pattern in the PV was not significantly altered in fetuses with restrictive FO. Therefore, assessment of PV flow did not improve the reliability of prediction.

Although we were able to correctly predict the postnatal FO anatomy with a sensitivity and specificity of 100% in our re-analysis, the neonates’ ability to achieve a sufficient O₂ saturation remains unpredictable. Among our 23 neonates with a normal FO size and normal FO flap mobility, 17% were ‘bad mixer’, who postnatally did not achieve a sufficient O₂ saturation and unexpectedly required urgent BAS also. As this situation escapes prenatal prediction, a normal FO anatomy consequently does not ensure an adequate interatrial shunting in all cases and the negative predictive value to reliably exclude an urgent BAS can naturally never reach 100%. This finding is in accordance with the study by Tuo et al. who described 10 of 40 fetuses that unexpectedly required urgent BAS (false-negative diagnoses in 25%) and who stated that “we should never trust fetuses with d-TGA, even when they seem to have a foramen ovale of normal appearance” [18]. Therefore, all fetuses with prenatally diagnosed d-TGA should always be delivered in a tertiary center with cardiac catheter stand-by allowing BAS within the first 24 h after birth, regardless of their predicted FO anatomy.

The prevalence of urgent BAS within the first 24 h in our cohort was 58%. In previously published prenatal series, the prevalence varied between 23% [15] and 57% [19]. These data are also comparable only to a certain extent because the indication to perform a BAS within the first 24 h was not standardized in different centers and depended on multiple factors including subjective assessment of first neonatal clinical presentation, variable institutional thresholds of oxygen saturation for urgent BAS, variable dosages of applied prostaglandin to prevent ductal closure, or the availability of a permanent cardiac catheter stand-by that allows urgent BAS at any time.

Besides the conventional echocardiographic assessment, very few recent studies described modern techniques on cardiac functional assessment to improve diagnostic accuracy and to optimize perinatal management [7, 24]. Patey and colleagues proposed a novel fetal index with cardiac parameters obtained by spectral pulsed wave tissue-Doppler imaging (PW-TDI) and speckle-tracking echocardiography (STE) and demonstrated significant cardiac geometrical and functional alterations in fetuses with d-TGA that required urgent BAS compared with fetuses that did not. They described the left ventricle rotation-to-shortening ratio, a ratio of apical systolic rotation to basal circumferential strain, as a marker of sub-endocardial dysfunction with the highest sensitivity for prediction of urgent BAS. However, this marker was only evaluated in a very small number of 13 pregnancies and has to be validated in larger prospective trials yet. So far it remains unclear, whether the use of PW-TDI or STE may also improve the prediction of the likelihood for urgent BAS in the subgroup of ‘bad mixer’ fetuses with normal FO size that escapes conventional echocardiography. In addition, calculation of the PW-TDI or STE parameters is more time-consuming, and the specificity of 83% seems to be no higher than the sensitivity of the previously published conventional echocardiographic parameters.

Other few series described the clinical utility of fetal cardiac magnetic resonance imaging (cardiac MRI) as a complementary tool to conventional echocardiography [23, 24, 28, 29]. Although MRI may improve fetal cardiovascular imaging, data on the use of cardiac MRI in prenatal diagnosis of cardiac defects including fetal d-TGA are still missing. Future studies are needed to provide more information on the reliability of fetal cardio MRI in the diagnosis of cardiac defects. Until then, conventional fetal echocardiography remains the ‘gold standard’ in fetal cardiology.

In fetuses with d-TGA, conventional fetal echocardiography with assessment of the FO size and FO flap mobility allows a reliable prenatal prediction of both, the restrictive and the normal FO, if fetal echocardiography is performed late in the third trimester within the last 2 weeks before birth. Prediction of the likelihood of urgent BAS also succeeds reliably in all fetuses with restrictive FO, but identification of the small subset of fetuses with normal FO anatomy that also requires urgent BAS (‘bad mixer’) fails, because the ability of sufficient postnatal interatrial mixing cannot be predicted prenatally. Therefore, all fetuses with prenatally diagnosed d-TGA should always be delivered in a tertiary center with cardiac catheter stand-by allowing BAS within the first 24 h after birth, regardless of their predicted FO anatomy.