* Correspondence: Professor M. Dommergues, Maternité, Hôpital Necker-Enfants Malades, 149 rue de Sèvres, 75743 Paris Cedex 15, France.
Objective To evaluate the potential of three-dimensional ultrasound to predict outcome in congenital diaphragmatic hernia.
Design Prospective observational study.
Setting Tertiary care centre.
Population Twelve cases of isolated congenital diaphragmatic hernia (11 left-sided, 1 right-sided) and 109 controls.
Methods Fetal lung volume was assessed by three-dimensional ultrasound using the technique of rotation of the multiplanar imaging. In the control fetuses, a logistic transformation was performed to correlate fetal lung volume with gestational age, and the confidence interval was obtained with a bootstrap resampling. A mathematical equation was then obtained allowing calculation of the expected fetal lung volume as a function of gestational age. In fetuses with congenital diaphragmatic hernia, the observed/expected lung volume ratio was compared with postnatal outcome.
Main outcome measures Neonatal mortality and pulmonary hypoplasia, which was defined as lung/body weight ratios less than 0.012.
Results The expected fetal lung volume was derived from the mathematical equation: Fetal lung volume (mL) = exp (4.72/(1 + exp ((20.32 − gestational age in weeks)/6.05))). The observed/expected fetal lung volume ratio was significantly lower in the congenital diaphragmatic hernia group (median: 0.34, range: 0.16–0.66), than in the control group (median: 1.02, range: 0.62–1.97, P < 0.0001). The distribution of this ratio was significantly downshifted in the infants with congenital diaphragmatic hernia who died (median: 0.19, range: 0.18–0.66) compared with survivors (median: 0.44, range: 0.36–0.66, P= 0.04). The observed/expected fetal lung volume ratio was also correlated with the postmortem lung/body weight ratio.
Conclusion In isolated congenital diaphragmatic hernia, fetal lung volume measurement by three-dimensional ultrasound is a potential predictor for pulmonary hypoplasia and postnatal outcome.
Congenital diaphragmatic hernia occurs with an incidence of approximately 1 in 2200 live births.1 Despite advances in antenatal and postnatal diagnosis and therapy, congenital diaphragmatic hernia is still associated with a high neonatal mortality rate, mainly due to pulmonary hypoplasia and pulmonary hypertension. Antenatal identification of fetuses with congenital diaphragmatic hernia who will die because of severe pulmonary hypoplasia at birth remains a challenge. Many prognostic factors based on conventional ultrasound (2DUS) have been evaluated such as left/right ventricle ratio,2,3 lung/head ratio,4,5 lung diameter/thoracic circumference ratio,6 liver position,7,8 amniotic fluid volume, mediastinal shift and stomach position.4,9 A prognostic score combining sonographic features has also been proposed.10 Recently, new potential prognostic factors were reported, such as fetal pulmonary artery diameters11 or the use of acceleration time/ejection time ratio in pulmonary arteries by Doppler blood flow velocimetry.12
Measuring the fetal lung volume in cases with congenital diaphragmatic hernia has been made possible using magnetic resonance imaging. While some authors have demonstrated a correlation between estimated fetal lung volume and postnatal outcome,13,14 others refuted the prognostic value of magnetic resonance imaging.15 Normal values of fetal lung volume have been established by three-dimensional ultrasound, but it has not so far been used in cases with congenital diaphragmatic hernia.16–20 We prospectively studied 12 fetuses with isolated congenital diaphragmatic hernia and 109 normal fetuses to evaluate the potential of fetal lung volume measurement by three-dimensional ultrasound and to predict outcome in cases with congenital diaphragmatic hernia.
From January to December 2002, a prospective study was conducted in our Fetal Medicine Unit. All women volunteered to undergo three-dimensional ultrasound imaging after being informed about the study protocol. Three-dimensional ultrasound was performed in 12 fetuses with a prenatal diagnosis of isolated congenital diaphragmatic hernia (11 left-sided and 1 right-sided) and in 109 controls. The gestational age was established by dates and by sonographic measurement of the crown–rump length in the first trimester.
The controls consisted of 109 normal fetuses, scanned at 20–37 weeks, in whom congenital abnormalities had been excluded by conventional ultrasound, which was then confirmed by subsequent neonatal examination. All controls were of normal birthweight. Fetuses with an abnormality that could possibly have induced lung hypoplasia, for example, oligohydramnios, were not enrolled as controls.
In the 12 cases with congenital diaphragmatic hernia, gestational age at diagnosis ranged from 16 to 35 weeks. A detailed ultrasound examination was performed. No associated anomaly was diagnosed, either antenatally or in the neonatal period. All fetuses had a normal karyotype. Three-dimensional ultrasound was performed between 29 and 36 weeks of gestation. The women were informed that the results of the fetal lung measurements would not be used to alter perinatal management. The perinatologists, ultrasonographers and surgeons in charge were not made aware of the results of the fetal lung volume measurements. The antenatal diagnostic investigations, obstetric care and postnatal treatment were provided in a single centre. In every case, immediately after birth the neonates were intubated and high frequency oscillatory ventilation commenced, which was subsequently change to conventional ventilation when deemed appropriate. In infants with severe pulmonary hypertension, nitric oxide was used (n= 7). Congenital diaphragmatic hernia repair was only performed after pre-operative stabilisation (n= 7). Five neonates with congenital diaphragmatic hernia survived, and seven died between the first day and the 22nd day of life. Among the babies that died, surgery was performed in two cases (cases one and three). When autopsy was performed the diagnosis of pulmonary hypoplasia was considered when the lung/body weight ratio was less than 0.012.21
A Voluson 730 ultrasound machine (Kretztechnik, Zipf, Austria), with a 4–8 MHz annular array transducer for three-dimensional volume scanning was used to evaluate fetal lung volume. For lung volume sampling, a transverse section of the fetal thorax at the level of the four-chamber view, with the fetal heart proximal to the transducer, was identified by conventional ultrasound and the volume box was adjusted to scan the entire fetal thorax. The maximum resolution was also adjusted and it took about 8 seconds to scan the whole volume. The angle of volume sampling varied throughout gestational age with a maximal limit of 75° in the third trimester. After scanning the volume, multiplanar imaging in the three orthogonal ultrasound sections was analysed to reconstruct the three-dimensional ultrasound image, which was stored on a removable hard disk. Before analysing their volume, each lung was carefully identified on the three orthogonal multiplanar imaging (Fig. 1). For this purpose, we had to take into account the anatomy of the pulmonary arteries, which were identified previously by two-dimensional power-Doppler ultrasound and also by working on the three-dimensional multiplanar imaging. This technique allowed us to distinguish the echogenicity of fetal lung from the herniated organs. Generally, the contralateral lung to the diaphragmatic defect was easier to identify.
In order to measure lung volume, a transverse section of the multiplanar imaging was chosen as the reference image for volume analyses. Right and left lung volumes were measured respectively by a series of area tracings of the studied lung after serial rotations through 30°. Reviewing parallel slices executed through the whole lung in the three perpendicular planes allowed us to check the adequacy of our rotational area tracing approach. This was made possible because the volume defined by rotational analysis could be reconstructed and displayed on the multiplanar imaging (Figs 2 and 3). Total lung volume, and left and right lung volumes were automatically measured and plotted against gestational age.
In the control group, total lung volume was log-transformed to correct for heterogeneity of the observed variance, which increased with gestational age. Different models of dependency were tested, including linear, exponential mode, gompertz and logistic models. The logistic mode was selected according to the minimum Akaike criterion. This yielded a formula allowing calculation of an expected total fetal lung volume for a given gestational age. A confidence interval of parameters was derived from bootstrap resampling method. All calculations were performed using R statistical software (http://cran.at.r-project.org/).
In the cases with congenital diaphragmatic hernia, fetal lung volume was expressed as the ratio of the measured observed lung volume to the expected fetal lung volume, based on the logistic transformation performed in controls. The mean observed/expected fetal lung volume in infants with congenital diaphragmatic hernia was compared with that of controls using a Mann–Whitney U test (StatView, Abacus, USA). Among the infants with congenital diaphragmatic hernia, the observed/expected fetal lung volume ratios of the survivors were compared with those of the non-survivors using the Mann–Whitney U test (StatView, Abacus).
All three-dimensional ultrasound examinations were performed by the same operator (R. R.). Intra-observer variability was assessed by analysing the measurements obtained from 10 controls, which were randomly selected, and from all cases of congenital diaphragmatic hernia (n= 12). For each pregnancy, fetal lung volume was estimated four times by the same operator. The intra-observer variability was calculated by analysis of variance (Excel, Microsoft, USA).
The following two-dimensional Doppler ultrasound prognostic factors were also analysed: (a) the presence of hydramnios; (b) major mediastinal shift when no fetal lung could be identified between heart and fetal thorax wall contralateral to the hernia on a four-chamber view; (c) liver herniation, considered significant when more than one-third of the liver appeared herniated, together with a kinked intra hepatic umbilical vein; (d) the presence of intrathoracic stomach; (e) lung surface/head diameter ratio; and (f) left ventricle/right ventricle ratio. The Mann–Whitney U test (StatView, Abacus) was used to compare quantitative parameters (lung surface/head diameter ratio and left ventricle/right ventricle ratio) and the χ2 test (Excel, Microsoft, USA) was used for qualitative parameters (hydramnios, major mediastinal shift, stomach location and liver herniation).
In the 109 controls, fetal lung volume increased as a function of gestational age, from an average of 9.95 mL at 20 weeks to 84.63 mL at 37 weeks (Fig. 4).
The logistic transformation analysis yielded the following formula:
where FLV represents the expected fetal lung volume and A, B and C correspond to unknown parameters to be estimated. The values of these parameters and their corresponding standard deviations (SD) are:
Figure 5 shows the box plots of the ratio of the observed/expected fetal lung volume for the controls and for the fetuses with congenital diaphragmatic hernia. The mean observed/expected fetal lung volume was significantly lower in the infants with congenital diaphragmatic hernia (median: 0.34, range: 0.16–0.66) than in controls (median: 1.02, range: 0.62–1.97; P < 0.0001). Among the infants with congenital diaphragmatic hernia, the observed/expected fetal lung volume ratio was significantly lower (median: 0.19, range: 0.16–0.66) in the seven cases resulting in neonatal death, than in the five survivors (median: 0.44, range: 0.36–0.66, P= 0.04).
The intra-observation variability was 1.46 mL for the controls and 0.66 mL for fetuses with congenital diaphragmatic hernia.
Six infants with congenital diaphragmatic hernia, whose observed/expected fetal lung volume ratio was lower than 0.35, died postnatally. Pulmonary hypoplasia was confirmed in the three cases in which the parents consented to a postmortem examination (Table 1).
Table 1. Fetal lung assessment by three-dimensional ultrasound and postnatal outcome.
Gestational age at diagnosis (weeks)
Gestational age at three-dimensional ultrasound (weeks)
Observed fetal lung volume (mL)
Observed/expected fetal lung volume
Surgery (days of life)
Lung weight (g)
Lung weight/ body weight
Right diaphragmatic hernia.
NND = neonatal death; A&W = alive and well; FIO2= minimal oxygen fraction; NO = minimal nitric oxide; NA = not available, no postmortem examination.
Among the six infants whose observed/expected fetal lung volume ratio were greater than or equal to 0.35, five (83.3%) survived and one (16.7%) died immediately following surgery because of acute pulmonary hypertension (case no. 7).
As for conventional ultrasound data, the presence of a major mediastinal shift (P= 0.02) and a lung surface/head diameter ratio lower than 2.0 (P= 0.04) was significantly associated with neonatal mortality. The other two-dimensional Doppler ultrasound findings were not significantly associated with outcome (Table 2).
Table 2. Prenatal conventional sonography. Mediastinal shift is considered as major when no fetal lung could be identified between heart and fetal thoracic wall contralateral to the hernia on a four-chamber view. Liver herniation is considered significant (Up) when more than one-third of the liver seemed herniated together with a kinked intrahepatic umbilical vein.
Our results show that fetal lung volume assessed by three-dimensional ultrasound is significantly lower in cases with congenital diaphragmatic hernia than in controls. This method may have a potential to predict neonatal outcome in congenital diaphragmatic hernia. Our findings are similar to those of previous reports using magnetic resonance imaging for evaluation of fetal lung volume in cases with congenital diaphragmatic hernia.13,14
In our study, fetal lung volume was estimated by the technique of rotation of the multiplanar imaging. We did not use the parallel slice method of the subtraction method for measuring the small and irregular lungs ipsilateral to the diaphragm defect. A normogram of fetal lung volume using the rotational was thus established. Fetal lung volume in controls varied according to gestational age (Fig. 3). The normogram established in the present study is similar to those obtained by magnetic resonance imaging.13,22 It is also similar to pioneer three-dimensional ultrasound studies16,18 based on subtracting the fetal heart volume from the fetal thoracic volume, at least from 28 to 35 weeks. Other three-dimensional ultrasound normograms using the subtraction approach tended to under-estimate fetal lung volume, although comparison with those data is made difficult by their relatively small size.17
Our estimates of normal lung volume were slightly larger than those of previous three-dimensional ultrasound studies measuring the lung volumes by the parallel slice technique.19,20 This discrepancy could be explained by the fact that the upper and the lower anatomical limits of the lung volume calculation using the parallel slice technique were set respectively at the level of the fetal clavicles and at the dome of the diaphragm. This may have led to under-estimating fetal lung volume as a small proportion of the lung is outside these limits.
The absence of previous reports on fetal lung volume assessment by three-dimensional ultrasound in cases with congenital diaphragmatic hernia could be explained by technical limitations. The only study including fetuses with possible pulmonary disturbance was recently reported by Osada et al.,23 but it included no case of congenital diaphragmatic hernia.
The previously used techniques for three-dimensional ultrasound fetal lung volume assessment might not be suitable for small lungs with distorted anatomy. Pioneer studies of three-dimensional ultrasound assessment of normal pulmonary volume, based on subtracting the fetal heart volume from the fetal thorax volume,16–18 cannot be used in congenital diaphragmatic hernia cases for evident reasons. In recent studies, fetal pulmonary volume was directly measured, based on outlining with a cursor, or a mouse, the contour of parallel slices of each lung. These studies provided normal pulmonary volumes calculated either from longitudinal or from transversal cross analyses.19,20 But as mentioned above, we failed to use this technique to estimate the volume of the fetal lung ipsilateral to the diaphragm defect.
Moreover, the technique of rotation of the multiplanar imaging seemed to be easier to perform and enabled us to afterwards confirm the volume on the three orthogonal plans of the multiplanar imaging. After finishing the volume calculation, the computer automatically outlined the limits of the fetal lung in each one of the three-orthogonal plans. This allows the operator to review these limits on parallel slices, and to correct them when necessary. Such an approach is particularly helpful when contouring small and irregularly shaped volumes, and probably explains the low intra-observer variability.
The significant association between the fetal lung volume ratio and neonatal mortality suggests the potential clinical use of three-dimensional ultrasound to predict outcome in cases with congenital diaphragmatic hernia. However, the size of the series is too small to compare the predictive value of three-dimensional ultrasound volume estimation with conventional ultrasound approaches, based for instance on the lung surface/head diameter ratio. We are aware that in our series lung surface/head diameter ratio and fetal lung volume measurements are similar predictors of neonatal death. Both indicators failed to foresee the demise due to acute pulmonary hypertension (case no. 7). This illustrates the complex pathophysiology of pulmonary lesions in congenital diaphragmatic hernia, which is not simply a functional problem of decreased lung volume but also involves vascular factors.
In conclusion, our results confirm that fetal lung volume in cases with congenital diaphragmatic hernia can be measured by three-dimensional ultrasound using the technique of rotation of the multiplanar imaging. This preliminary study suggests that low fetal lung volume estimated by three-dimensional ultrasound in fetuses with congenital diaphragmatic hernia may be associated with pulmonary hypoplasia and neonatal mortality. But the small size of the study does not allow definitive conclusions to be drawn regarding the prognostic value of three-dimensional ultrasound and larger studies are necessary to assess the clinical value of three-dimensional ultrasound in fetuses with congenital diaphragmatic hernia.