Objective
The purpose of this study was to evaluate observed/expected (O/E) lung-to-head ratio (LHR) by ultrasound (US) and total fetal lung volume (TFLV) by magnetic resonance imaging as neonatal outcome predictors in isolated fetal congenital diaphragmatic hernia (CDH).
Study Design
We conducted a retrospective study of 72 fetuses with isolated CDH, in whom O/E LHR and TFLV were evaluated as survival predictors.
Results
O/E LHR on US and O/E TFLV by magnetic resonance imaging were significantly lower in newborn infants with isolated CDH who died compared with survivors (30.3 ± 8.3 vs 44.2 ± 14.2; P < .0001 for O/E LHR; 21.9 ± 6.3 vs 41.5 ± 17.6; P = .001 for O/E TFLV). Area under receiver-operator characteristics curve for survival for O/E LHR was 0.80 (95% confidence interval, 0.70–0.90). On multivariate analysis, O/E LHR predicted survival, whereas hernia side and first neonatal pH did not. For each unit increase in O/E LHR, mortality odds decreased by 11% (95% confidence interval, 4–17%).
Conclusion
In fetuses with isolated CDH, O/E LHR (US) independently predicts survival and may predict severity, allowing management to be optimized.
Survival rates of 60-80% have been reported for infants with isolated congenital diaphragmatic hernia (CDH). Prenatal diagnosis and advancements in neonatal care have led to improved survival, but there remains a significant risk of morbidity in survivors, mainly because of lung hypoplasia and resultant pulmonary hypertension. The prediction of severity of pulmonary hypoplasia and thus postnatal survival relies on imaging modalities that indirectly assess fetal lung volume. The most commonly used method is the measurement of lung area to head circumference ratio (LHR) with the use of 2-dimensional ultrasound (US). Many studies have reported improved survival with increasing LHR; however, it has been observed that LHR changes with advancing gestational age; therefore, a fixed cutoff, as initially proposed, can be misleading. To counteract the variations in lung and head measurements at different gestations, an observed-to-expected (O/E) LHR has been suggested.
Recently, total fetal lung volume (TFLV) measurements by magnetic resonance imaging (MRI) have also been reported to be useful in the prediction of pulmonary hypoplasia. Some studies report a significantly higher likelihood of death if the O/E TFLV were <35%. Some correlation has been observed between US- and MRI-derived estimates, and it has been suggested that the prognostic accuracy of O/E TFLV and O/E LHR by MRI may be slightly better than that of sonographic O/E LHR. The side of CDH and liver herniation have been reported as additional prognostic markers.
Antenatal detection of pulmonary hypoplasia not only predicts neonatal survival, but also may help in the selection of fetuses (those considered least likely to survive with conventional postnatal therapy) for fetal intervention by fetal endotracheal occlusion to promote fetal lung growth. Our aim was to evaluate O/E LHR that had been measured by 2-dimensional US and O/E TFLV that had been measured by MRI as predictors of neonatal death and morbidity for fetuses with prenatally diagnosed isolated CDH.
Materials and Methods
We conducted a retrospective study of all cases of isolated prenatally diagnosed CDH from 18-38 weeks of gestation that were referred to the Fetal Medicine Unit at Mount Sinai Hospital, Toronto, Canada, from January 2002 to April 2010. Fetuses with associated major malformations, chromosomal abnormalities, bilateral CDH, fetuses who delivered <30 weeks’ gestation, and pregnancies which were terminated were excluded from our analysis. Approval for this study was obtained from the research ethics boards at both institutions.
Digitally stored images from the detailed anatomic US at the first visit to the Fetal Medicine Unit were reviewed. The best images with a transverse section of the fetal chest, which contained a 4-chamber view of the heart, were chosen to measure LHR. The area of the contralateral lung was measured by the longest axis method, as described by Metkus et al. This involved multiplication of the longest diameter of the lung by its longest perpendicular diameter ( Figure 1 ). All US measurements were taken from 18-38 weeks of gestation. The LHR measurements were obtained retrospectively, if they had not been recorded prospectively. This was performed independently by 2 observers (M.A.A. and G.R.) who were blinded to outcome. After 2005, the LHR measurements were obtained prospectively; however, these results were never used when patients were counseled regarding outcome. The observed LHR was then expressed as a percentage of the expected mean for gestational age (expected LHR: left = –0.0042 × [gestational age in weeks] 2 + 0.3043 × gestational age – 2.5957; right = –0.0048 × [gestational age in weeks] 2 + 0.3995 × gestational age – 3.4802), as proposed by Jani et al.
MRI (1.5T GE Signa; GE Healthcare, Waukesha, WI) was performed with predominantly fast spin echo sequences. Fetal lung volume (FLV) was performed by the tracing of a region of interest around individual consecutive images of lung tissue on fast spin echo sequences images. The volumes for both lungs were obtained and added for TFLV ( Figure 2 ). The expected TFLV by MRI was determined with the following formula by Rypens et al : expected TFLV= 0.0033 × (gestation in weeks) 2,86 which was expressed as a percentage of O/E TFLV. All MRIs were done from 23-34 weeks of gestation. All MRIs were read prospectively by 1 radiologist who was blinded to both the O/E LHR measurements derived from US and to the postnatal outcome. Data, if available, that included gestational age, head circumference, side of hernia, and liver position/herniation were collected from USs and MRIs.
After birth, these children were stabilized initially at our perinatal center. Infants were intubated and ventilated with the aid of sedation and neuromuscular blockade. Positive pressure ventilation was initiated, with the aim of the prevention of any ventilation-induced lung injury by avoidance of high peak inspiratory pressures. The objective of ventilation was the establishment of a satisfactory preductal arterial oxygen saturation (>85%), while hypercarbia (PaCO 2 , 45-55 mm Hg) was tolerated, if necessary, as long the pH remained >7.35. Bicarbonate was used to correct for pH values below that level when appropriate. In the event of hypercarbia or the need for high peak airway pressures, high frequency jet ventilation was used. Neonates were transferred to the pediatric intensive care unit in the adjacent Hospital for Sick Children for further treatment.
The protocol for stabilization before repair included the use of high frequency oscillation ventilation at low mean airway pressures (<16 cm H 2 O) to minimize pulmonary barotrauma and echocardiography to assess the degree of pulmonary hypertension and the status of the ductus arteriosus. Additional therapy included intravenous prostaglandin to maintain ductal patency and inhaled nitric oxide to reduce pulmonary artery pressures. Extra corporeal membrane oxygenation (ECMO) was used in those infants with suprasystemic pulmonary artery pressures, but without severe pulmonary hypoplasia, in whom a preductal oxygen saturation of >85% could not be maintained. Surgical repair was deferred until each infant’s condition was stabilized on conventional mechanical ventilation. These postnatal therapeutic protocols were in place for the entire duration of the study.
The following outcome parameters were collected for each infant: survival to discharge, birthweight, Apgar scores at 1 and 5 minutes, umbilical arterial pH, neonatal resuscitation details, first neonatal arterial blood gas, ECMO use, the duration of assisted ventilation, oxygen supplementation, and length of hospitalization.
Data analysis
Baseline characteristics and outcomes were compared between survivors and nonsurvivors by χ 2 or Fisher’s exact tests for categoric variables and t test or Wilcoxon rank sum test for continuous variables. The interobserver reliability of O/E LHR by US was assessed by an intraclass correlation coefficient. To facilitate the examination of the association between survival and O/E LHR, the study cohort was divided into 5 groups according to O/E LHR by US (≤25%, 26–35%, 36–45%, 46–55%, >55%) and 4 groups according to O/E TFLV by MRI (≤25%, 26–35%, 36–45%, ≥46%). The Cochran-Armitage trend test was used to examine the survival trends across the groups. In addition, multiple logistic regression models were used to investigate the effect of O/E LHR on survival, which was adjusted for the risk factors and potential confounders identified in the univariate analysis. A receiver-operator characteristics curve was constructed, and the area under the curve (AUC) was estimated to determine the accuracy of O/E LHR by US in the prediction of postnatal survival. The accuracy or power of each prognostic variable in the prediction of survival was assessed with the AUC, sensitivity, and specificity information. Statistical software (version 9.2; SAS Institute Inc, Cary, NC) and R 2.10.1 ( www.r-project.org ) were used for the analyses.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
