Objective
Congenital diaphragmatic hernia (CDH) results in morbidity and death from lung hypoplasia and persistent pulmonary hypertension (PH). We sought to define the relationship between fetal ultrasound markers of severity in CDH and the time to resolution of neonatal PH.
Study Design
We conducted a retrospective study of fetuses with an antenatal ultrasound scan and left-sided CDH cared for at the University of California San Francisco (2002-2012). Fetal liver position was classified on ultrasound scan as abdominal (entire liver within the abdomen) or thoracic (any portion of the liver within the thorax). Fetal stomach position was classified from least to most aberrant: abdominal, anterior left chest, mid-posterior left chest, or retrocardiac (right chest). Lung-to-head ratio (LHR) was determined from available scans at 20-29 weeks of gestational age (GA). Routine neonatal echocardiograms were performed weekly for up to 6 weeks or until PH resolved or until discharge. PH was assessed by echocardiogram with the use of a hierarchy of ductus arteriosus level shunt, interventricular septal position, and tricuspid regurgitant jet velocity. Days to PH-free survival was defined as the age at which pulmonary artery pressure was estimated to be <2/3 systemic blood pressure. Cox proportional hazards models adjusted for GA at birth, era of birth, fetal surgery, and GA at ultrasound scan (LHR model only), with censoring at 100 days.
Results
Of 118 patients, the following fetal markers were available: LHR (n = 53), liver position (n = 112), and stomach position (n = 80). Fewer infants experienced resolved PH if they had LHR <1 ( P = .006), thoracic liver position ( P = .001), or more aberrant stomach position ( P < .001). There was also a decreased rate of resolution of PH in infants with LHR <1 (hazard ratio, 0.30; P = .007), thoracic liver position (hazard ratio, 0.38; P < .001), and more aberrant stomach position (hazard ratios, 0.28 [ P = .002]; 0.1 [ P < .001]; and 0.07 [ P < .001]).
Conclusion
Fetal ultrasound markers of CDH severity are predictive not only of death but also of significant morbidity. LHR <1, thoracic liver, and aberrant stomach position are associated with delayed time to resolution of PH in infants with CDH and may be used to identify fetuses at high risk of persistent PH.
Congenital diaphragmatic hernia (CDH) occurs in 1:4000-5000 live births and is associated with significant morbidity and death. Contemporary care strategies have improved the survival of these infants dramatically, particularly at referral centers. Although this improvement is substantial, many of these infants are surviving with significant morbidity related to their pulmonary hypoplasia and pulmonary hypertension (PH) that leads to a prolonged need for assisted ventilation, home oxygen use, rehospitalization, developmental delay, and complicated medication use.
Despite the high rates of morbidity, research in this population has focused heavily thus far on only death and the need for extracorporeal membrane oxygenation (ECMO). As such, antenatal sonographic markers of severity in fetuses with CDH have been used primarily to predict death and the need for ECMO and to select candidates for fetal intervention. The lung-to-head ratio (LHR), particularly LHR <1, and thoracic liver position (as compared with abdominal position) have been shown to be predictors of both death and the need for ECMO use. More recently, it has been demonstrated that increasingly abnormal stomach position is a strong predictor of not only death and ECMO but also a need for prolonged neonatal respiratory support. This is likely due to the relationship of stomach position to liver position.
We and others have shown that persistence of PH at >2-3 weeks of life is associated with poor neonatal outcomes; thus, there may be benefits to early identification of these infants who are at high risk of morbidity for antenatal counseling, selection for fetal intervention, and targeting future early neonatal therapies. This makes the prediction of persistent PH in infants with CDH an important and useful determination during fetal life. This study is the first to assess the ability of fetal LHR, liver position, and stomach position to predict both resolution of PH and the time required to achieve its resolution.
Materials and Methods
Patient data
We performed a retrospective cohort study of fetuses with antenatal ultrasound scans and left-sided Bochdalek-type CDH at our center, who subsequently were cared for after birth at the University of California San Francisco Benioff Children’s Hospital (2002-12). Fetuses were included if they had ≥1 neonatal echocardiogram and excluded if they had multiple congenital anomalies or a known or suspected syndrome. Clinical data were collected by chart review. Echocardiograms were rereviewed by a reader who was blinded to clinical data. The outcome of interest was days to resolution of PH, defined as the chronologic age at which the echocardiogram estimated <2/3 systemic systolic pulmonary arterial pressure. Chronic lung disease was defined as the requirement for any respiratory support at 56 days of age or home oxygen if discharged earlier than 56 days. Patients were treated with gentle ventilation and lung-sparing strategies, as previously described. The Institutional Review Board of the University of California San Francisco approved this study.
Imaging and echocardiographic data
Fetal sonographic markers of interest included the LHR, liver position, and stomach position. LHR measurements were gathered from ultrasound reports or, if absent from the report, were measured directly from stored images, when available, by a single radiologist who was blinded to the clinical outcome. LHR was included in the analysis only if the measurement was taken from a sonogram that had been performed at 20-29 weeks gestational age (GA). All lung dimensions were measured on a transaxial view of the fetal chest at the level of the 4-chamber view of the heart. The technique consistently used and reported at University of California, San Francisco, uses the aorta and lateral rib as landmarks for the lateral measurement, with the orthogonal anterior-posterior diameter measured from the cardiac atria to the posterior rib, then dividing by the head circumference. All analyses were done first including scans from 22-27 weeks’ GA and then expanding to 20-29 weeks’ GA. Because results did not differ when we used these 2 inclusion criteria, data from the broader range of ages are used to increase the information provided in our analyses. Liver position was gathered from fetal ultrasound reports and verified by operative report from CDH repair, if surgery was performed. Six infants, all with intrathoracic liver on fetal sonography, died before surgical repair; thus, liver position could not be verified. Liver position was categorized as abdominal (defined as entire liver within the abdomen) or intrathoracic (defined as any portion of the liver herniated into the thoracic cavity).
A single radiologist who was blinded to neonatal outcomes reviewed the initial fetal sonogram that was performed at our institution to determine stomach position when adequate images were available. Stomach position was classified based on the degree of herniation into the thoracic cavity while the fetal thorax was viewed in the true axial plane at the level of the 4-chamber view of the heart. Positions included abdominal, anterior left chest (defined as a portion of the fetal stomach that contacted the anterior chest wall), mid-to-posterior left chest (defined as not contacting the anterior left chest wall and either contacting or possibly contacting the posterior left chest wall), or retrocardiac (defined as at least a portion of the stomach located posterior to the left atrium of the heart within the right chest).
Echocardiogram protocol
Beginning in 2002, all infants with CDH underwent echocardiograms (C256 and C512 and SC2000; Acuson Sequoia, Mountain View, CA), per routine clinical protocol, within the first 48 hours of life and then weekly for the first 6 weeks of life or until resolution of PH, death, or discharge. Standard clinical views were obtained to estimate the degree of elevation of pulmonary arterial pressure relative to systemic systolic blood pressure. Echocardiograms were classified into 1 of 2 categories: <2/3 systemic systolic pressures (no/mild PH) and ≥2/3 systemic systolic pressures (PH). Classifications were made with a hierarchy of measurements that involved (1) direction and velocity of flow via the ductus arteriosus, (2) interventricular septum position, and (3) peak tricuspid regurgitant jet velocity, by the modified Bernoulli equation and with the assumption of right atrial pressure as 0 mmHg, as previously described.
Statistical analysis
Between groups univariate comparisons were made with χ 2 , Student t test, Mann Whitney rank-sum, Kruskal Wallis, and analysis of variance tests where appropriate. Kaplan Meier curves were compared with the use of Log-rank tests. Cox proportional hazards multivariate models were created to assess the relationship between fetal markers of CDH severity and time to resolution of PH. Infants were censored at 100 days; infants who died were treated as being alive and never having resolved their PH and were censored at 100 days of age to avoid informative censoring. We adjusted for GA at birth, era of birth (2002-2005, 2006-2008, and 2009-2012), history of fetal tracheal occlusion, and, for the LHR model only, GA at sonography. Finally, a Harrell’s C statistic was generated for each model to assess the predictive accuracy of each fetal marker in predicting time to resolution of PH (Stata 12.0 software; Stata Corporation, College Station, TX).
Results
We identified 118 infants with left-sided CDH who had fetal sonography at University of California, San Francisco. Cohort characteristics are described in Table 1 . The cohort was predominantly term, with an overall mortality rate of 19% (23/118 infants). Seventy-three percent (86/118 infants) experienced resolution of their PH before death or discharge ( Figure 1 ), with a median time to resolution, among those whose PH resolved, of 14 days (interquartile range, 9–20). LHR was available in 89 patients (75%), 53 of which were obtained from 20-29 weeks’ GA and included in the final analysis. There were 29 infants in whom LHR was not noted in the fetal ultrasound report and for whom original images were unavailable for primary review. Of the 118 infants, 80 (68%) had adequate images available for determination of stomach position. Liver position was documented in 112 fetuses. The distribution of these measurements is shown in Table 2 . There were no obvious differences in newborn characteristics between subcohorts, as determined by fetal measurement ( Table 1 ). Additionally, there was no difference in proportion of subjects with available measurements by era across the 10-year study period for liver position, stomach position, or LHR and thus no suggestion of bias introduced by differential missing data by era.
| Neonatal characteristics | All infants (n = 118) | Lung-to-head ratio available a (n = 53) | Liver position available (n = 112) | Stomach position available (n = 80) |
|---|---|---|---|---|
| Gestational age, wk b | 38 ± 2.3 | 37 ± 2.6 | 38 ± 2.3 | 38 ± 2.2 |
| Male sex, % (n) | 58 (68) | 53 (28) | 59 (66) | 55 (44) |
| Fetal tracheal occlusion surgery, % (n) | 8 (9) | 15 (8) | 8 (9) | 10 (8) |
| Year of birth, % (n) | ||||
| 2002-2005 | 30 (35) | 26 (14) | 29 (33) | 29 (23) |
| 2006-2008 | 25 (29) | 23 (12) | 25 (28) | 25 (20) |
| 2009-2012 | 46 (54) | 51 (27) | 46 (51) | 46 (37) |
| Time to resolution of pulmonary hypertension, d c d | 14 (9–20) | 15 (10–21) | 14 (9–18) | 15 (9–21) |
| Death, % (n) | 19 (23) | 32 (17) | 21 (23) | 25 (20) |
| Extracorporeal membrane oxygenation, % (n) | 13 (15) | 15 (8) | 13 (15) | 15 (12) |
| Chronic lung disease, % (n) | 23 (22) | 28 (10) | 22 (20) | 28 (17) |
a Patients included only if fetal ultrasound scan was performed between 20 and 29 weeks of gestational age: 3 infants were excluded for scan at <20 weeks of gestation, and 33 infants were excluded for ultrasound scan at >29 weeks of gestation
c Time to estimated pulmonary arterial pressures <2/3 systemic systolic pressures among those whose condition resolved before death/discharge
| Ultrasound parameter | Measure |
|---|---|
| Lung-to-head ratio (n = 53) a | 1.2 ± 0.5 |
| Gestational age at ultrasound scan, wk a | 24 ± 2.4 |
| Lung-to-head ratio <1, % (n) | 36 (19) |
| Liver position (n = 112), % (n) | |
| Abdominal | 46 (52) |
| Intrathoracic | 54 (60) |
| Stomach position (n = 80), % (n) | |
| Abdominal | 16 (13) |
| Anterior left chest | 23 (18) |
| Mid-to-posterior left chest | 43 (34) |
| Retrocardiac (right chest) | 19 (15) |
Infants with an LHR <1 had increased mortality rates before discharge (58% vs 18%; P = .001), increased need for ECMO (32% vs 6%; P = .012), and increased incidence of chronic lung disease (75% vs 14%; P = .001) compared with infants with an LHR ≥1. Additionally, a lower proportion of infants with LHR <1 had resolution of their PH before death or discharge (42% vs 76%; P = .012; Figure 2 ) and a longer time to resolution of PH among those who ultimately experienced resolution (median, 22 vs 14 days; P = .003). After adjustment for GA at birth and GA at the time of fetal ultrasound scan, fetal tracheal occlusion, and era of birth, infants with LHR <1 continued to have a prolonged time to resolution ( Table 3 ). However, LHR is only a moderate predictor of rate of resolution of PH, with a C statistic of 0.68.
| Variable | Survival to discharge, n (%) | Experienced resolution of pulmonary hypertension before death or discharge, n (%) | Time to resolution of PH, d a,b | Adjusted hazard ratio c,d | 95% confidence interval | P value |
|---|---|---|---|---|---|---|
| Lung-to-head ratio (n = 53) | ||||||
| ≥1 (n = 34) | 28 (82) | 26 (76) | 14 (9–15) | Referent | Referent | – |
| <1 (n = 19) | 8 (42) | 8 (42) | 22 (18–30) | 0.3 | 0.13–0.72 | .007 |
| Liver position (n = 112) | ||||||
| Abdominal (n = 52) | 51 (98) | 46 (88) | 11 (8–15) | Referent | Referent | – |
| Intrathoracic (n = 60) | 38 (63) | 36 (60) | 17 (12–28) | 0.38 | 0.24–0.61 | < .001 |
| Stomach position (n = 80) | ||||||
| Abdominal (n = 13) | 13 (100) | 13 (100) | 9 (6–12) | Referent | Referent | – |
| Anterior left chest (n = 18) | 17 (94) | 17 (94) | 15 (14–16) | 0.28 | 0.13–0.62 | .002 |
| Mid-to-posterior left chest (n = 34) | 24 (71) | 21 (62) | 17 (9–22) | 0.1 | 0.04–0.23 | < .001 |
| Retrocardiac (n = 15) | 6 (40) | 7 (47) | 22 (11–34) | 0.07 | 0.02–0.20 | < .001 |
a Time to estimated pulmonary arterial pressures <2/3 systemic systolic pressures among those who resolved before death/discharge
b Data are given as median (interquartile range)
c After adjustment for gestational age, history of fetal surgery, birth era, and, for lung-to-head ratio model, gestational age at fetal ultrasound scan with the use of Cox proportional hazards models
d Lower hazard ratios represent a decreased chance of resolution of PH.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
