Machine Learning to Predict Outcomes of Fetal Cardiac Disease: A Pilot Study.

Prediction of outcomes following a prenatal diagnosis of congenital heart disease (CHD) is challenging. Machine learning (ML) algorithms may be used to reduce clinical uncertainty and improve prognostic accuracy. We performed a pilot study to train ML algorithms to predict postnatal outcomes based on clinical data. Specific objectives were to predict (1) in utero or neonatal death, (2) high-acuity neonatal care and (3) favorable outcomes. We included all fetuses with cardiac disease at Sunnybrook Health Sciences Centre, Toronto, Canada, from 2012 to 2021. Prediction models were created using the XgBoost algorithm (tree-based) with fivefold cross-validation. Among 211 cases of fetal cardiac disease, 61 were excluded (39 terminations, 21 lost to follow-up, 1 isolated arrhythmia), leaving a cohort of 150 fetuses. Fifteen (10%) demised (10 neonates) and 65 (48%) of live births required high acuity neonatal care. Of those with clinical follow-up, 60/87 (69%) had a favorable outcome. Prediction models for fetal or neonatal death, high acuity neonatal care and favorable outcome had AUCs of 0.76, 0.84 and 0.73, respectively. The most important predictors for death were the presence of non-cardiac abnormalities combined with more severe CHD. High acuity of postnatal care was predicted by anti Ro antibody and more severe CHD. Favorable outcome was most predicted by no right heart disease combined with genetic abnormalities, and maternal medications. Prediction models using ML provide good discrimination of key prenatal and postnatal outcomes among fetuses with congenital heart disease.

Safety and feasibility pilot study of continuous low-dose maternal supplemental oxygen in fetal single ventricle heart disease.

OBJECTIVES: Fetuses with single ventricle physiology (SVP) exhibit reductions in fetal cerebral oxygenation with associated delays in fetal brain growth and neurodevelopmental outcomes. Maternal supplemental oxygen (MSO) has been proposed to improve fetal brain growth but current evidence on dosing, candidacy, and outcomes are limited. In this pilot study, we evaluated the safety and feasibility of continuous low-dose MSO in the setting of SVP. METHODS: This single-centre, open-label, pilot phase 1 safety and feasibility clinical trial included 25 pregnant individuals with a fetal diagnosis of SVP. Participants self-administered continuous supplemental oxygen using medical-grade oxygen concentrators for up to 24 hours per day from the second half of gestation until delivery. The primary aim was the evaluation of the safety profile and feasibility of MSO. A secondary preliminary analysis was performed to assess the impact of MSO on the fetal circulation by echocardiography and late-gestational cardiovascular magnetic resonance, early outcomes including brain growth and pre-operative brain injury, and 18-month neurodevelopmental outcomes by the Bayley Scales of Infant and Toddler Development 3rd Edition compared to a contemporary fetal SVP cohort that received standard of care (SOC). RESULTS: Among 25 participants, the average maternal age at conception was 35 years, and fetal SVP diagnoses included 16 right ventricle dominant, 8 left ventricle dominant, and 1 indeterminant ventricular morphology. Participants started the trial at approximately 29.3 gestational weeks and took MSO for a median 16.1 hours per day for 63 days, cumulating a median 1029 hours of oxygen intake from enrollment until delivery. The only treatment-associated adverse events were nasal complications that were typically resolved by attaching a humidifier unit to the oxygen concentrator. No premature closure of the ductus arteriosus or unexpected fetal demise was observed. In the secondary analysis, MSO was not associated with any changes in fetal growth, middle cerebral artery pulsatility index, cerebroplacental ratio, nor head circumference to abdominal circumference ratio Z-scores over gestation compared to SOC. Although MSO was associated with changes in umbilical artery pulsatility index Z-score over gestation compared to SOC (p=0.02), this was likely due to initial baseline differences in placental resistance. At late-gestational cardiovascular magnetic resonance, MSO was not associated with any significant increase in umbilical vein oxygen saturation, fetal oxygen delivery, or fetal cerebral oxygen delivery. Similarly, we observed no differences in newborn outcomes including brain volume and pre-operative brain injury, nor mortality by 18 months of age, nor neurodevelopmental outcomes at 18 months of age. CONCLUSIONS: This pilot phase 1 clinical trial indicates low-dose maternal supplemental oxygen therapy is safe and well tolerated in pregnancies diagnosed with fetal SVP. However, our protocol was not associated with any significant changes in fetal circulatory physiology or improvements in early neurologic or neurodevelopmental outcomes. This article is protected by copyright. All rights reserved.

Fetal cardiac findings and hemodynamic changes associated with severe lower urinary tract obstruction in utero.

OBJECTIVES: To describe fetal echocardiographic findings associated with lower urinary tract obstruction (LUTO) and to compare anatomic and hemodynamic measurements between fetuses with LUTO and gestational age (GA)-matched controls, with an emphasis on quantitative indices of diastolic function and cardiac output. METHODS: This was a retrospective cohort study of fetuses diagnosed with severe LUTO with giant bladder, which underwent at least one fetal echocardiogram at our center between January 2005 and June 2018. Fetuses with major congenital heart disease were excluded. Control fetuses did not have any structural or functional abnormalities and were GA-matched to the LUTO fetuses based on the time of the first fetal echocardiogram. Cardiac anatomy and hemodynamic measurements were compared between fetuses with LUTO and controls. In infants with LUTO, serial fetal and postnatal echocardiographic data were assessed, when available, and clinical outcomes were reviewed. RESULTS: Twenty-six fetuses with LUTO and at least one fetal echocardiogram available were identified, one of which was excluded due to hypoplastic left heart syndrome, leaving 25 LUTO fetuses in the final cohort. The mean GA at the first fetal echocardiogram was 25.4 ± 5.1 weeks in the LUTO group and 25.3 ± 5.0 weeks in the control group. Common findings in fetuses with LUTO included cardiomegaly (40%), pericardial effusion (44%), right ventricular (RV) hypertrophy (64%) and left ventricular (LV) hypertrophy (48%). Compared with GA-matched controls, LUTO fetuses had lower ascending aorta Z-score (-0.10 ± 0.94 vs -0.93 ± 1.03; P = 0.02) and aortic isthmus Z-score (-0.14 ± 0.86 vs -1.62 ± 1.11; P < 0.001), shorter mitral valve inflow time indexed to cardiac cycle length (0.46 ± 0.04 vs 0.41 ± 0.06; P = 0.002), and worse (increased) LV myocardial performance index (0.39 ± 0.03 vs 0.44 ± 0.04; P < 0.001). In addition, the ratio of RV to LV cardiac index was higher in LUTO fetuses compared with controls (1.62 ± 0.13 vs 1.33 ± 0.11; P < 0.001). Of the 25 LUTO pregnancies, two were lost to follow-up, three underwent elective termination of pregnancy and three ended in intrauterine fetal demise. Four (16%) patients had mildly hypoplastic left-heart structures, comprising two with aortic arch hypoplasia and two with mitral and aortic stenosis. CONCLUSION: In addition to presenting with cardiomegaly, pericardial effusion and ventricular hypertrophy, fetuses with LUTO demonstrate LV diastolic dysfunction and appear to redistribute cardiac output as compared to control fetuses, which may contribute to the development of left-heart hypoplasia. Copyright © 2019 ISUOG. Published by John Wiley & Sons Ltd.

Myocardial strain abnormalities in fetuses with pulmonary atresia and intact ventricular septum.

OBJECTIVES: Global and regional myocardial deformation have not been well described in fetuses with pulmonary atresia and intact ventricular septum (PA/IVS). Speckle-tracking echocardiography (STE), an angle-independent technique for assessing global and regional strain, may be a more sensitive way of determining ventricular systolic dysfunction compared with traditional 2D echocardiography. The aim of this study was to assess myocardial deformation in fetuses with PA/IVS compared with control fetuses and to determine if, in fetuses with PA/IVS, strain differs between those with and those without right ventricle-dependent coronary circulation (RVDCC). METHODS: This was a retrospective analysis of fetuses with PA/IVS examined at two medical centers between June 2005 and October 2017. Left ventricular (LV) and right ventricular (RV) regional and global longitudinal strain (GLS) and strain rate were obtained using STE, and comparisons were made between fetuses with PA/IVS and gestational age (GA)-matched controls. Postnatal outcome was assessed, including the presence of RVDCC. RESULTS: Fifty-seven fetuses with PA/IVS and 57 controls were analyzed at a mean GA of 26.5 ± 5 weeks. LV-GLS was significantly decreased in fetuses with PA/IVS compared with controls (-17.4 ± 1.7% vs -23.7 ± 2.0%, P < 0.001). LV strain rate was also significantly decreased (-1.01 ± 0.21/s vs -1.42 ± 0.20/s, P < 0.001). Fetuses with PA/IVS had decreased strain in all segments. Similarly, RV strain was significantly decreased in fetuses with PA/IVS (-11.6 ± 3.8% vs -24.6 ± 2.5%, P < 0.0001). Thirty-six patients had postnatal cardiac catheterization performed to define coronary anatomy; 10 fetuses had RVDCC. Fetuses with RVDCC had decreased LV strain compared with those without (-15.8 ± 1.2% vs -17.9 ± 1.7%, P = 0.009). RV strain was also decreased in fetuses with RVDCC vs those without (-7.0 ± 2.9% vs -12.1 ± 3.2%, P = 0.0004). CONCLUSIONS: Fetuses with PA/IVS have decreased global and regional LV and RV strain compared with controls. The finding of decreased LV strain may be due to altered ventricular mechanics in the context of a hypertensive right ventricle and/or abnormal coronary perfusion. Moreover, fetuses that were found to have RVDCC postnatally had decreased LV and RV strain compared with those that did not. These results encourage further investigation to assess whether fetal ventricular strain could be a prenatal predictor of RVDCC. Copyright © 2018 ISUOG. Published by John Wiley & Sons Ltd.

Improved technical success, postnatal outcome and refined predictors of outcome for fetal aortic valvuloplasty.

OBJECTIVES: Fetal aortic valvuloplasty (FAV) may prevent progression of mid-gestation aortic stenosis to hypoplastic left heart syndrome (HLHS). The aim of this study was to evaluate whether technical success and biventricular (Biv) outcome after FAV have changed from an earlier (2000-2008) to a more recent (2009-2015) era and identify pre-FAV predictors of Biv outcome. METHODS: We evaluated procedural and postnatal outcomes in 123 fetuses that underwent FAV for evolving HLHS at Boston Children's Hospital between 2000 and 2015. The primary outcome measure was circulation type (Biv vs single ventricle) at the time of neonatal hospital discharge. Classification and regression tree (CART) analysis was performed to construct a stratification algorithm to predict Biv circulation based on pre-FAV fetal variables. RESULTS: The FAV procedure was technically successful in 101/123 (82%) fetuses, with a higher technical success rate in the more recent era than in the earlier one (49/52 (94%) vs 52/71 (73%); P = 0.003). In liveborn patients, the incidence of Biv outcome was higher in the recent than in the earlier era, both in the entire liveborn cohort (29/49 (59%) vs 16/62 (26%); P = 0.001) and in those in whom the procedure was technically successful (27/46 (59%) vs 15/47 (32%); P = 0.007). Independent predictors of Biv outcome were higher left ventricular (LV) pressure, larger ascending aorta, better LV diastolic function and higher LV long-axis Z-score. On CART analysis, fetuses with LV pressure > 47 mmHg and ascending aorta Z-score ≥ 0.57 had a 92% probability of Biv outcome (n = 24). Those with a lower LV pressure, or mitral dimension Z-score < 0.1 and mitral valve inflow time Z-score < -2 (n = 34) were unlikely to have Biv (probability of 9%). The remainder of the patients had an intermediate (∼40-60%) likelihood of Biv circulation. CONCLUSIONS: The proportion of patients achieving Biv outcome after FAV has increased, probably owing to an improved technical success rate and modified selection criteria. Fetal factors, including LV pressure, size of the ascending aorta and diastolic function, are associated with likelihood of Biv circulation after FAV. Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd.

Low rate of prenatal diagnosis among neonates with critical aortic stenosis: insight into the natural history in utero.

OBJECTIVES: To better understand the natural history and spectrum of fetal aortic stenosis (AS), we aimed to (1) determine the prenatal diagnosis rate of neonates with critical AS and a biventricular (BV) outcome, and (2) describe the findings at fetal echocardiography in patients diagnosed prenatally. METHODS: A multicenter, retrospective study was performed on neonates who presented with critical AS and who were discharged with a BV outcome from 2000 to 2013. The prenatal diagnosis rate was compared with that reported for hypoplastic left heart syndrome (HLHS). We reviewed fetal echocardiographic findings in patients who were diagnosed prenatally. RESULTS: In only 10 (8.5%) of 117 neonates with critical AS and a BV outcome was the diagnosis made prenatally, a rate significantly lower than that for HLHS in the contemporary era (82%; P < 0.0001). Of the 10 patients diagnosed prenatally, all had developed left ventricular dysfunction by a median gestational age of 33 (range, 28-35) weeks. When present, Doppler abnormalities such as retrograde flow in the aortic arch (n = 2), monophasic mitral inflow (n = 3) and left-to-right flow across the foramen ovale (n = 8) developed late in gestation (median 33 weeks). CONCLUSION: The prenatal diagnosis rate of critical AS and a BV outcome among neonates is very low, probably owing to a relatively normal four-chamber view in mid-gestation with development of significant obstruction in the third trimester. The natural history contrasts with that of severe mid-gestation AS with evolving HLHS and suggests that the gestational timing of development of significant AS has an important impact on subsequent left-heart growth in utero.

Prenatal diagnosis of transposition of the great arteries over a 20-year period: improved but imperfect.

OBJECTIVE: To evaluate temporal trends in the prenatal diagnosis of transposition of the great arteries with intact ventricular septum (TGA/IVS) and its impact on neonatal morbidity and mortality. METHODS: We included in this study cohort newborns with TGA/IVS who were referred for surgical management to our center over a 20-year period (1992-2011). The study period was divided into five 4-year periods and the primary outcome was rate of prenatal diagnosis. Secondary outcomes included neonatal preoperative status and perioperative survival. RESULTS: Of the 340 patients with TGA/IVS, 81 (23.8%) had a prenatal diagnosis. The rate of prenatal diagnosis increased over the study period, from 6% in 1992-1995 to 41% in 2008-2011 (P < 0.001). Compared to patients with a postnatal diagnosis, balloon atrial septostomy (BAS) was performed earlier in patients with a prenatal diagnosis (0 days after delivery vs 1 day after delivery, respectively; P < 0.001) and fewer prenatally diagnosed neonates required mechanical ventilation (55.6% vs 68.0%; P = 0.03). Between patients with a prenatal or postnatal diagnosis of TGA/IVS, there were no statistically significant differences in the incidence of preoperative acidosis (16.0% vs 25.5%; P = 0.1), need for preoperative extracorporeal membrane oxygenation (2.5% vs 2.7%; P = 1.0) or mortality (one preoperative and no postoperative deaths among prenatally diagnosed patients compared with four preoperative and six postoperative deaths among postnatally diagnosed patients). CONCLUSIONS: The prenatal detection rate of TGA/IVS has improved but still remains below 50%, suggesting the need for strategies to increase detection rates. The mortality rate was not statistically significantly different between prenatally and postnatally diagnosed patients, however, there were significant preoperative differences with regard to earlier BAS and fewer neonates that required mechanical ventilation. Ongoing work is required to ascertain whether prenatal diagnosis confers long-term benefits.