Fetal Intervention for Congenital Heart Disease

In 1982, Harrison et al published a letter in the New England Journal of Medicine asserting that there were certain "simple [fetal] structural defects... whose alleviation might allow fetal development to proceed normally." At that time, the "simple structural defects" amenable to intervention included congenital diagphragmatic hernia, hydronephrosis, and hydrocephalus. In the years since that landmark paper, however—in large part as a consequence of technical and scientific advances in maternal-fetal medicine, surgical technique, and fetal echocardiography—fetal therapy has also grown to include fetal cardiac intervention (FCI). 

Experimental animal models of open fetal cardiac surgery began in the 1980s, with the aim of describing the physiologic and pathologic impacts of extracorporeal circulatory bypass. ^[1]  Although initial research showed promise, the fetoplacental response to bypass was characterized by cytokine activation, endothelial dysfunction, and increased resistance in the placenta, with fetal hypoxia and demise as end results. ^{[2, 3, 4, 5, 6, 7, 8]}  As a result, open fetal cardiac surgery was largely abandoned.

The earliest reported human fetal cardiac therapy of any kind took place in 1975 and involved maternal-fetal transplacental administration of a beta blocker in the setting of fetal ventricular tachyarrhythmia. ^[9]  The first open in-utero cardiac procedure was reported a decade later, in 1986, with a pacemaker placement for complete heart block. ^[10]  The concept of performing balloon valvuloplasty in fetuses with stenotic heart valves followed the successful introduction of neonatal balloon valvuloplasty in the 1980s, with the first reported case performed in a fetus with aortic stenosis (AS) in 1991. ^[11]

Fetal cardiac intervention in current era

Fetal therapy is a broad term that encompasses a range of transplacental medications, catheter-based interventions, fetoscopic procedures, minimally invasive fetoscopic surgical procedures, open fetal surgical procedures, and ex-utero intrapartum treatment (EXIT) procedures. FCI, the focus of this article, is a term referring to catheter-based procedures for a narrow subset of congenital heart defects. ^[12] At present, these defects include the following ^{[13, 14, 15]} :

• Severe midgestation AS with evolving hypoplastic left heart syndrome (eHLHS),
• Pulmonary atresia with intact ventricular septum (PA/IVS) and evolving hypoplastic right heart syndrome (eHRHS)
• Hypoplastic left heart syndrome with an intact or highly restrictive atrial septum (HLHS/RAS) ^[16]

Rationale

Severe midgestation AS with eHLHS

Although the majority of cardiac structures have formed by week 10 of gestation, a percentage of patients with severe midgestation AS are known to progress to the full complex of HLHS by the third trimester. ^{[17, 18, 19]} The fetal myocardium is particularly sensitive to changes in afterload; thus, increasing degrees of left ventricular (LV) outflow tract (LVOT) obstruction are poorly tolerated and may result in ventricular dysfunction, myocardial damage, and eventual growth arrest (see the videos below). ^{[20, 17, 18, 21]}

The goal of fetal aortic valvuloplasty (FAV) is to relieve discrete valvular obstruction before irreversable myocardial injury is incurred, thereby promoting antegrade blood flow and growth of left-heart structures and increasing the chances of achieving a biventricular circulation after birth. ^{[22, 23]}

HLHS can evolve from a simple semilunar valve stenosis (see the videos above). 

PA/IVS and eHRHS

PA with intact or highly restrictive atrial septum exists along a wide clinical spectrum. At one extreme, there is fibromuscular atresia of the pulmonary valve, a markedly hypoplastic and muscle-bound right ventricle (RV), and coronary artery anomalies, necessitating a univentricular strategy after birth. ^{[24, 25, 26]}  On the milder end of the spectrum, there are cases of membranous PA with a normal-sized right ventricle, which are likely to achieve a biventricular circulation. Fetal pulmonary valvuloplasty (FPV) would not meaningfully change the clinical course for either of these two groups.

For the cases that fall between these two extremes, the clinical trajectory is much less clear. Whereas overall rates of right-heart structures have been described in this disease process, ^[26]  there is significant variation, such that two fetuses with identical measurements at midgestation may have different prospects for a biventricular circulation at birth. The goal of FPV is to promote antegrade blood flow through the right heart, thus encouraging growth and function of these structures and increasing the chances of a biventricular circulation. ^{[27, 28, 29]}

HLHS/RAS

It is well established that a severely restrictive or intact atrial septum confers a significantly worse prognosis in HLHS, with an early mortality of 50-90%  ^[30] . Atrial-level restriction results in left atrial hypertension, with resulant pulmonary vascular changes that include muscularization of the pulmonary veins and lymphangectasia ("nutmeg lung"). These maladaptive morphologic changes may be irreversible and are considered a particularly high-risk feature along the single-ventricle pathway. ^{[31, 32, 33]}  

Beyond in-utero lung damage, RAS is also associated with significant morbidity and mortality in the immediate postnatal period. After the neonate takes the first breath, the sudden drop in pulmonary vascular resistance results in a marked increase in pulmonary blood flow. In the setting of RAS, there is no mechanism for left atrial egress, and severe left atrial hypertension and pulmonary edema result.

Without an immediate intervention, acidosis, cyanosis, and demise may progress within a matter of minutes. Measures aimed at rapidly decompressing the left atrium, such as balloon atrial septostomy, are considered particularly high-risk, in part because long-standing left atrial hypertension can promote marked thickening of the interatrial septum. ^{[34, 30, 35, 36]}

Because the time between delivery and hemodynamic decompensation may be quite short, the diagnosis of HLHS/RAS has historically required herculean efforts to coordinate delivery. The details of management are center-dependent, but generally, such efforts may include geographic relocation of the mother before delivery, scheduling a cesarean section during weekday hours, and coordination of catheterization and surgical staff for emergency intervention. ^{[37, 38]}

Fetal intervention for RAS was proposed to decompress the left atrium as a means of providing hemodynamic stability during delivery and the immediate postnatal period. This hemodynamic stability may also enable mothers to have a more "normal" delivery experience, including the potential for a vaginal delivery. ^[39]

Severe midgestation AS with eHLHS

Patient selection

Because a rigorous approach to patient selection is critical for maximizing procedural success, criteria for candidacy for intervention continue to be carefully refined as experience evolves. ^{[40, 41]}  Patient selection is a multidisciplinary decision that must consider a number of variables, including the fetal cardiac substrate, fetal comordities, gestational age, and maternal health.

In broad terms, candidacy requires the appropriate substrate of valvar AS, hemodynamic markers that predict progression to HLHS, ^[20]  and the right timing (ie, before the LV has incurred irreversible damage).

The dominant cardiac anatomy is valvar AS with all of the following features:

• Decreased mobility of valve leaflets
• A narrow color jet across the LVOT (diameter of antegrade color jet < diameter of aortic anulus)
• No or minimal subvalvar obstruction

Hemodynamic markers predicting progression to HLHS include the following:

• Depressed LV function
• Either retrograde or bidirectional flow in the transverse aortic arch or two of the following: (1) monophasic mitral inflow Doppler pattern, (2) left-to-right flow across atrial septum or intact atrial septum, (3) bidirectional flow in pulmonary veins

The following considerations are applied to determining LV salvageability:

• LV long-axis Z-score should be greater than zero; the presence of endocardial fibroelastosis (EFE) or a short, fat, globular LV indicates that myocardial insult and growth arrest have already occurred 
• Mitral regurgitation or aortic outflow peak systolic gradient should be greater than 20 mm Hg, preferably higher; although some ventricular dysfunction is expected, advanced ventricular dysfunction suggests that myocardial damage has already occurred, and this has been associated with poor outcomes 

McElhinney et al examined a consecutive Boston cohort (2000-2008) and devised a scoring system optimized for 100% sensitivity for predicting a biventricular outcome. ^[42]  This scoring system relies on the following five variables:

• LV long-axis Z-score greater than 0 (1 point)
• LV short-axis Z-score greater than 0 (1 point)
• Aortic anulus Z-score greater than –3.5 (1 point)
• Mitral valve anulus Z-score greater than –2 (1 point)
• Mitral regurgitation or aortic outflow peak systolic gradient greater than 20 mm Hg (1 point)

A composite score of 4 or higher had 100% sensitivity and 53% specificity for identifying fetuses that had a biventricular outcome after birth.

Friedman et al compared consecutive Boston cohorts (2000-2008 and 2009-2015) and found that the later cohort had higher rates of technical success (94% vs 73%) and biventricular circulation among liveborn patients (59% vs 26%). ^[41] These improvements were largely attributed to refinements in patient selection—specifically, exclusion of candidates with low LV pressure or left-heart stuctures below Z-score thresholds.

The authors performed classification and regression tree (CART) analysis to determine fetal echocardiographic characteristics with the greatest power to identify biventricular outcome. ^[41] In this analysis, the variables with the highest discriminating power for biventricular outcome were the following:

• LV pressure - Whereas prior publications used 20 mm Hg as a lower threshold for candidacy, this model demonstrated that the likelihood of a biventricular outcome was low (10-20%) in patients with an LV pressure between 20 and 30 mm Hg but increased linearly with increasing LV pressure; the likelihood of a biventricular outcome reached 60% at an LV pressure of 50 mm Hg
• Ascending aorta Z-score
• LV long-axis Z-score
• Mitral valve inflow time Z-score 

Timing

To prevent further myocardial damage, once the above criteria are met, the intervention should be arranged as soon as possible. The ideal gestational age for treatment is between 18 and 30 weeks' gestation.

PA/IVS and eHRHS

Patient selection

In a 2020 collaborative paper from the International Fetal Cardiac Intervention Registry (IFCIR), criteria for patient selection appeared to vary substantially. ^[43]  The following criteria are derived from single-center experiences ^{[44, 27]} :

• Membranous pulmonary atresia with the following features: identifiable pulmonary valve leaflets, no valve opening in systole, and no color flow across the valve
• Intact or highly restrictive ventricular septum
• Right-heart hypoplasia - Tricuspid valve anulus Z-score of –2 or less; qualitatively small but identifiably tripartite RV

Timing

In PA/IVS, as compared with fetal aortic stenosis, the RV may have more capacity to remodel (on the basis of observations in postnatal patients) and likely does not incur as much myocardial damage in utero, for unknown reasons. Intervention has been performed between 21 and 28 weeks' gestation, though anecdotal evidence suggests improved technical success after 24 weeks. ^[44]

HLHS/RAS

Patient selection

At present, there are no published guidelines on what degree of atrial restriction warrants intervention. Criteria are center-dependent but may reasonably include the following ^{[45, 46, 47]} : 

• Small atrial communication on two-dimensional (2D) imaging (foramen ovale ≤ 1 mm)
• Pulmonary vein forward-to-reverse velocity-time integral (VTI) ratio less than 3 (ratios between 3 and 5 may also be considered) ^{[34, 31, 32, 33]}
• Secondary findings of pulmonary lymphangiectasia on magnetic resonance imaging (MRI)

Timing

Optimal timing for these interventions has not been studied. Earlier intervention offers the benefit of reducing morbidity caused by long-standing left atrial hypertension but probably poses a greater risk of procedural complications. In a 2017 IFCIR report, most of these procedures were performed between 25 and 32 weeks' gestation. ^[39]

Contraindications for FCI include the following:

• Significant preexisting maternal disease or obstetric comorbidity that would place the fetus or mother at higher risk - Relative contraindications include body mass index (BMI) greater than 35, communicable diseases such as HIV infection, uncontrolled or pregestational diabetes, history of cervical incompetence, and hematologic disorders that affect coagulation
• Significant extracardiac pathology in the fetus, including structural anomalies and chromosomal abnormalities - Noninvasive maternal serum testing should be performed (when available) to confirm euploidy; invasive confirmation of a normal karyotype or microarray (eg, amniocentesis or chorionic villus sampling) is center-dependent ^[48]
• Multiple gestation (relative)
• Inability of the pregnant person to provide informed consent

Fetal aortic valvuloplasty

The first reported successful series of per-ventricular FAVs to halt the evolution of HLHS in fetuses with AS from a single center was published by Tworetzky et al in 2004. ^[49]  Of the 20 pregnancies in which this procedure was performed, 14 (70%) cases were technically successful, with 21% of fetuses subsequently achieving biventricular circulation after birth.

Larger dimensions of left-heart structures and higher LV pressure have been retrospectively recognized as predictors of successful FAV and eventual biventricular circulation, whereas moderate-to-severe EFE at the time of the procedure is associated with lack of response despite technically successful valvuloplasty. ^{[42, 50]}  (See the images below.)

Left atrial decompression in HLHS/RAS

Two approaches to left atrial decompression have been described: atrial septoplasty and atrial septal stent placement. In one series, of 19 successful cases after fetal atrial septoplasty, 12 required additional procedures in the immediate postnatal period. ^[45]  Atrial septal stent placement has been proposed as a more durable mechanism for left atrial egress. Belfort et al made further refinements to their approach with introduction of a thulium laser to vaporize a hole in the thickened interatrial septum. ^[35]

Certain anatomic features can complicate this procedure. In some cases, the interatrial septum is markedly thickened and irregular (see the video below). In others, the left atrium is particularly small. This latter situation may be considered a contraindication if the only angle of approach is through the large right atrium, because of the concern for an insufficient "landing zone" for the stent on the left atrial side. Cor triatriatum is considered a contraindication for this procedure. 

Fetal pulmonary valvuloplasty

A study from the IFCIR described a cumulative procedural success rate of 71% across institutions. In the initial series from the Boston group, the first four procedures reported were technical failures; however, the subsequent six procedures were technically successful, reflecting a steep learning curve for the procedure. (See the video below.)

Fetal aortic valvuloplasty

In multiple series, FAV has been shown to increase the odds of a biventricular circulation. In a study by Freud et al that evaluated the first 100 fetuses who had undergone FAV for severe midgestation AS with eHLHS in Boston (median follow-up, 5.4 y), 38 of the 88 (39%) who were born alive had a biventricular circulation, either from birth or after initial univentricular palliation. ^[51]

Another group from Linz, Austria, reported that in as many as 43% of fetal AS cases, successful antenatal intervention led to biventricular circulation after birth, with 67% of live-born technical successes being biventricular. ^[52] Other factors (eg, older gestational age at the time of intervention and, perhaps, less severe LV dysfunction at the time of the procedure) may have contributed to the higher success rates seen in this series in comparison with the initial Boston series.

Although achieving a biventricular circulation is an admirable goal that holds promise in reducing morbidity associated with single-ventricle palliation, current data on morbidity, quality of life, and long-term survival are lacking. ^{[53, 54, 55, 56, 57]} Preliminary data from from the IFCIR showed that overall 2-year survival was similar between those who received FAV and eligible fetuses who did not.

The Boston group published a decision-tree model that projected survival over 6 years. When comparing projected FAV survival based on contemporary data to survival as a single ventricle (extrapolated from the single-ventricle reconstruction trial), they found that the FAV approach could offer a survival benefit (though not a statistically significant one), as long as fetal demise from the procedure remained below 12%.

Initial data suggested that neurodevelopmental outcomes are similar to those of patients with HLHS. ^[58]

Overall, it is clear from present studies that in-utero balloon valvuloplasty as an isolated procedure has not obviated the need for additional postnatal intervention, including repeat aortic valvuloplasty, repair of coarctation, resection of EFE, mitral valvuloplasty, and temporary left atrial decompression procedures. ^{[59, 60, 52, 45, 61, 42]}

Fetal pulmonary valvuloplasty

Most outcome data for FPV have come from single-center studies. A publication from IFCIR represented the first multicenter description of dyad characteristics, complications, and neonatal outcomes. ^[43] Fetuses who underwent FCI were compared with fetuses who were offered but did not undergo the procedure. Because criteria for intervention varied between centers and the baseline tricuspid valve dimensions and degree of tricuspid valve regurgitation were greater in the FCI group, statistical comparisons were not performed, and the results were descriptive. 

Right-heart growth and biventricular outcome

With the baseline differences between the groups acknowledged, the absolute dimensions and Z-scores of tricuspid valve dimensions grew at a greater rate among those who underwent successful FPV. ^[43] Whereas the frequency of a biventricular outcome was greater in the FPV group, 60% of interventions were performed on fetuses with baseline tricuspid valve Z-scores greater than –2. Data suggest that fetuses with this relatively mild degree of right-heart hypoplasia are likely to achieve a biventricular circulation regardless.  

Complications and survival

Fetal complications occurred in 55% of cases, with pericardial effusion necessitating pericardiocentesis (48%) and bradycardia (36%) being the most common. ^[43] There were seven periprocedural deaths and two additional deaths occurring more than 48 hours after the procedure (both attributed to circular shunt physiology), for an overall mortality of 15.5% in the FPV group. In comparison, 100% of the nonintervention group survived to delivery. Eighteen percent of the fetuses were delivered prematurely (30-36.6 wk), with FPV cases accounting for 73% of them. 

Left atrial decompression for HLHS/RAS

At present, limited outcome data are available for this procedure. ^[62] A report from the IFCIR described preliminary outcomes from their collaborative effort. ^[39]  In all, 47 interventions from eight participating centers were evaluated, including 27 septoplasties and 20 atrial septal stents. Comparisons drawn between the two groups did not reach statistical significance, but there were several notable trends. There appeared to be a greater chance of technical success for atrial septoplasty (85%) than for atrial stent placement (65%), but in the successful cases, stents tended to be more reliable than septoplasties in maintaining a nonrestrictive foramen ovale at the time of delivery (75% vs 39%).

In fetuses who received an intervention, overall survival to hospital discharge was poor (35%) but was similar to that in FCI candidates who did not undergo the procedure, likely reflecting the known dangers of this high-risk anatomic substrate. ^[39] Among those who underwent successful stent placement, survival to discharge was 58%. Moreover, in cases characterized by both procedural success and a nonrestrictive foramen ovale at birth, 1-year survival was 59%. In contrast, among the cohort that was offered but did not undergo FCI, 1-year survival was 19%. 

Additional studies are required to determine patient eligibility, ideal timing of intervention, risks associated with repeat interventions after initial procedural failure, and longer-term morbidity and mortality.