Intervention for fetal critical aortic stenosis: first step in a metamorphosis?

Percutaneous ultrasound-guided fetal cardiac intervention is increasingly used to alter the prognosis of specific forms of congenital heart disease. Usually, this is just the first step in a complex chain of procedures. One such intervention is fetal aortic valvuloplasty (FAV) for critical aortic stenosis, the subject of studies of Friedman et al.1 and Kovacevic et al.2 in this issue of the Journal. Critical aortic stenosis in neonates is associated with varying degrees of underdevelopment of left heart structures that may impact on postnatal management and outcome. When there is only mild left ventricular (LV) hypoplasia, a biventricular (BV) outcome is usual, although, after relief of neonatal aortic stenosis, most cases require later additional interventions, including replacement of the aortic valve3. Severe hypoplasia of the left heart (hypoplastic left heart syndrome, HLHS) renders it unable to support the systemic circulation, mandating either staged single-ventricular palliation or cardiac transplantation. Between these two extremes is the group with moderate left heart hypoplasia (‘borderline LV’), in whom it is often unclear whether relief of aortic stenosis will suffice to allow the left heart to support the systemic circulation. It is well-established that some fetuses diagnosed with critical aortic stenosis and a normal or dilated LV during the second trimester subsequently progress to HLHS4-6. The primary aim of FAV is to prevent HLHS, if this is the predicted outcome in the absence of intervention, and to potentially enable BV outcome7. Although FAV has shown promise in averting HLHS, many of these children are born with a borderline LV and a spectrum of anomalies, notably endocardial fibroelastosis and mitral valve hypoplasia/dysplasia. This pathology may further impair cardiac performance and impede BV repair. Neonatal decision-making regarding the optimal management of borderline LV is a challenge, as the decision regarding whether to pursue a univentricular or BV circulation is often made very soon after birth, is not easily reversed and may prove fatal. Significant diastolic dysfunction and pulmonary hypertension is not infrequent following BV repair8 and, with hindsight, these cases may have been better served by univentricular repair9. Conversely, staged single-ventricular palliation is associated with significant morbidity and mortality, with only a minority of HLHS patients surviving to adulthood without a cardiac transplant10-12. This explains the desire to attempt BV circulation whenever feasible. An alternative approach defers the ultimate decision regarding ventricular outcome to later in infancy, by bilateral banding of the pulmonary arteries and maintaining ductal patency13. This allows more time to determine the adequacy of the mitral valve, the magnitude of any postnatal left heart growth and LV diastolic dysfunction14. Using postnatal primary BV circulation after mid-gestational FAV as a measure of procedural success, there are important differences between the two articles in this issue1, 2. Boston Children's Hospital has the most experience with FAV and their patient selection criteria have been refined repeatedly over the last 20 years15, 16. In this update by Friedman et al.1, all 123 FAV cases were grouped into one of two eras, 2000–2008 (n = 71) and 2009–2015 (n = 52). Following revision of selection criteria in 200916, FAV outcomes improved, with technical success of 94% vs 73%, procedure-related in-utero mortality of 4% vs 8% and primary BV outcome of 59% vs 26%. Six additional cases underwent staged BV repair. In the European study by Kovacevic et al.2, the proportion of 67 cases undergoing FAV at one of six European sites between 2005 and 2012 which achieved a BV outcome was 36% and none underwent staged BV repair. This outcome was not different from that of a previously published ‘natural history’ cohort without FAV (BV outcome: 38%) that included cases from additional European centers17. There are various prenatal and postnatal factors that may explain differences in outcome among FAV cohorts. Mid-gestational critical aortic stenosis is a rare antenatal diagnosis18 and even the largest programs will, on average, perform only five to 10 FAV procedures annually. There is a learning curve with any procedure, and skills, techniques and results improve with experience19. Recent BV outcome data from larger single centers1, 19-21 tend to be better than pooled results from multicenter studies that include many smaller programs2, 22. This suggests that a critical case load may be required for optimal results. Institutional experience, consistency of approach and case volume may be even more important in the postnatal than in the fetal management of patients with borderline left heart hypoplasia, as there are often strong institutional preferences towards either defaulting to a univentricular pathway or pursuing a potentially complex, multistage approach to left heart rehabilitation23. Closer examination of the European study's2 supporting information (table S1) reveals that one center contributed half (33/67) of all FAV cases, attempted significantly more BV repairs (16/29 (55%) vs 4/18 (22%) operated cases; odds ratio, 4.31; 95% CI, 1.14–16.30; P = 0.04) and reported higher postoperative survival rates (83% vs 66%) compared with the other European sites. Centers with rates of BV outcome lower than expected based on ‘natural history’ data17 should review critically their pre- and postnatal management strategies. Undoubtedly, further improvements are required in the accuracy of predicted outcomes for critical fetal aortic stenosis to avoid unnecessary in-utero procedures. Friedman et al.1 placed patients into three subgroups, according to whether they had ‘high’, ‘intermediate’ or ‘low or no’ likelihood of BV outcome following FAV, based on anatomical (ascending aortic and mitral valve Z-scores) and functional (LV pressure, mitral inflow) parameters. Nevertheless, a major limitation of all predictive models to date includes the lack of adjustments for gestational age or fetal weight. FAV has been performed as early as 19 and as late as 35 gestational weeks19-22, 24. In figure 2 of Friedman et al.1 (univariate modeling results), the likelihood of BV outcome following FAV is about 20% at 20 weeks but 70% at 30 weeks. Several studies have shown that successful FAV leads to improved but subnormal fetal left heart growth2, 16, 24. The concept that the earlier that critical aortic stenosis occurs, the greater the growth failure and the smaller will be the left heart structures at term, is clinically important2, 16. FAV in a 20-week fetus is associated with an increased risk of procedure-related fetal demise and little chance of BV outcome, irrespective of the presence of ‘favorable’ echocardiographic findings. On the other hand, a late-gestation fetus has a low risk of procedure-related in-utero demise but there is questionable benefit from the intervention as survival with a BV outcome may be achieved without in-utero intervention1. Future predictive models of ventricular outcome after FAV should incorporate gestational age and address at which gestational age an intervention becomes irrelevant to BV circulation. In conclusion, based on data from the two largest-volume centers in two different studies, 55–59% of severe fetal aortic stenosis cases that undergo balloon dilation attain BV repair, which is an improvement on BV repair rates in natural history cohorts. However, the fetal management is the simplest aspect of this process, as there are myriad postnatal decisions and procedures to navigate to achieve BV repair. Any beneficial effect of FAV in terms of improved left heart growth or function may be masked by the complexity of the postnatal management and requires institutional commitment to optimize outcome.