European Radiology

, Volume 23, Issue 5, pp 1299–1305

Liver-to-thoracic volume ratio: use at MR imaging to predict postnatal survival in fetuses with isolated congenital diaphragmatic hernia with or without prenatal tracheal occlusion

Authors

  • Mieke M. Cannie
    • Department of RadiologyUniversity Hospital Brugmann
    • Department of RadiologyUZ Brussel, Vrije Universiteit Brussel
  • Anne-Gaël Cordier
    • Department of Obstetrics and Gynecology and Centre de Maladie Rare: Hernie de Coupole DiaphragmatiqueHôpital Antoine Béclère, APHP, Université Paris Sud
  • Jocelyne De Laveaucoupet
    • Department of RadiologyHôpital Antoine Béclère, APHP, Université Paris Sud
  • Stéphanie Franchi-Abella
    • Department of Pediatric RadiologyHôpital Kremlin Bicètre, APHP, Université Paris Sud
  • Maud Cagneaux
    • Department of Pediatric and Fœtal RadiologyHôpital Femme Mère Enfant, Université Claude Bernard Lyon I
  • Olivier Prodhomme
    • Department of Pediatric RadiologyMontpellier University Hospital
  • Marie-Victoire Senat
    • Department of Obstetrics and GynecologyHôpital Kremlin Bicètre, APHP, Université Paris Sud
  • Mostafa Mokhtari
    • Department of NeonatologyHôpital Kremlin Bicètre, APHP, Université Paris Sud
  • Vinciane Vlieghe
    • Department of NeonatologyHôpital Universitaire des Enfants Reine Fabiola
  • Dorota Nowakowska
    • Department of Fetal-Maternal Medicine and GynecologyMedical University and the Research Institute Polish Mother’s Memorial Hospital
  • Alexandra Benachi
    • Department of Obstetrics and Gynecology and Centre de Maladie Rare: Hernie de Coupole DiaphragmatiqueHôpital Antoine Béclère, APHP, Université Paris Sud
    • Department of Obstetrics and GynecologyFetal Medicine and Treatment Unit, University Hospital Brugmann
Pediatric

DOI: 10.1007/s00330-012-2709-6

Cite this article as:
Cannie, M.M., Cordier, A., De Laveaucoupet, J. et al. Eur Radiol (2013) 23: 1299. doi:10.1007/s00330-012-2709-6

Abstract

Objective

To evaluate the relationship of the liver-to-thoracic volume ratio (LiTR) by MRI with postnatal survival in foetuses with isolated congenital diaphragmatic hernia (CDH).

Methods

In 30 conservatively managed CDH foetuses and in 31 who underwent fetoscopic endoluminal tracheal occlusion (FETO), logistic regression analysis was used to investigate the effect on postnatal survival of the observed-to-expected (O/E) ratio of total foetal lung volume (TFLV), LiTR, gestational age at delivery, CDH side, intrathoracic position of the liver and, for those who underwent FETO, gestational age at FETO and occlusion period. For 19 foetuses undergoing FETO, a post-FETO MRI was available. The proportionate increase in O/E ratio of TFLV at 3–8 weeks after FETO was compared with the pre-FETO value and correlated with pre-FETO LiTR using linear regression analysis.

Results

For conservatively managed foetuses, only LiTR provided a significant prediction of postnatal survival. For foetuses undergoing FETO, LiTR and gestational age at delivery provided a significant independent prediction of postnatal survival. There was a significant inverse association between lung response and pre-FETO LiTR.

Conclusion

In foetuses with CDH with/without FETO treatment, the LiTR is predictive of postnatal survival at discharge. In foetuses treated with FETO, LiTR is predictive of post-FETO lung response.

Key Points

Congenital diaphragmatic hernia is usually managed conservatively before surgery soon after delivery

Fetoscopic endoluminal tracheal occlusion (FETO) has been introduced for severely affected foetuses

In conservatively managed CDH, the liver-to-thoracic volume ratio (LiTR) predicted postnatal survival best.

In severe CDH with prenatal FETO, LiTR also helped predict postnatal survival.

LiTR should be integrated into the prenatal decision-making for foetuses with CDH.

Keywords

Liver-to-thoracic volume ratioDiaphragmatic herniaMagnetic resonance imagingTracheal occlusionLung response

Introduction

Congenital diaphragmatic hernia (CDH), with a mean incidence of 1 in 3,000 births, can be diagnosed in the prenatal period either as part of other anomalies or as an isolated birth defect [1]. It is potentially a lethal birth defect involving herniation of abdominal contents into the thoracic cavity through a defect in the diaphragm. Up to 30 % of babies with isolated CDH still die in the neonatal period from the consequences of lung hypoplasia and/or pulmonary hypertension [28].

Congenital diaphragmatic hernia is one of the rare congenital diseases that could be potentially amenable to prenatal endoscopic surgery. Prenatal surgery in CDH consists of inserting a balloon into the foetal trachea in order to promote lung growth, a technique known as fetoscopic endoluminal tracheal occlusion (FETO) [911]. Although minimally invasive, FETO remains associated with a relatively high risk of preterm birth [11]. It is therefore essential that cases are correctly selected for eligibility to prenatal treatment.

Traditionally, criteria for FETO are based on prenatal ultrasound measurement of the observed-to-expected (O/E) ratio of the lung-to-head ratio (LHR) and the intrathoracic position of the liver. Eligible cases are those classified as severe such as left-sided CDH with intrathoracic herniation of the liver and O/E ratio of LHR <25 % [12] and right-sided CDH [13]. Furthermore, moderate cases such as left-sided CDH with intrathoracic liver herniation and an O/E ratio of LHR up to 44.9 % are currently also offered FETO aiming to decrease postnatal morbidity [14, ClinicalTrials.gov Identifier NCT00763737].

Recently, a magnetic resonance (MR) imaging study including 40 foetuses with CDH who underwent FETO has shown that, independently of the pre-FETO O/E ratio of total foetal lung volume (TFLV), measurement of lung volume responsiveness evaluated at 2 to 5 weeks post-FETO was predictive of postnatal survival [15]. In another MR imaging study including 40 conservatively managed CDH foetuses, the liver-to-thorax volume ratio (LiTR), being a measure of the amount of herniated liver into the thoracic cavity, was shown to be, independently from the O/E ratio of TFLV, predictive of postnatal survival [16]. To date, the predictive value of LiTR in CDH foetuses undergoing FETO has to the best of our knowledge never been described.

The aim of our study was to evaluate LiTR measured by means of MR imaging in foetuses with isolated CDH who were conservatively managed or who underwent FETO in the prediction of postnatal survival.

Materials and methods

Study participants and design

The study was approved by the local ethics committee, and all patients gave written informed consent to the FETO procedure. For the retrospective collection of MR imaging data, written informed consent was not required. This was a multicentre retrospective, cross-sectional study conducted at the Departments of Radiology and Foetal Medicine Units of the University Hospital Brugmann; Brussels; Belgium, from May 2008 to March 2012. Participants included pregnant women with foetuses with isolated CDH who were either managed conservatively until delivery, or, when severe or moderate, were offered FETO. Criteria for FETO are based, as traditionally performed, on prenatal ultrasound measurement of the O/E ratio of LHR [1214]. The FETO procedure was performed using local or loco-regional anaesthesia as previously described [13]. The balloon was inserted into the foetal trachea at around 25–30 weeks of gestation and we aimed to remove the balloon at 34 weeks of gestation. The balloon was removed earlier in case of premature rupture of the membranes and threatened premature birth. Inclusion criteria for this study were as follows: (1) availability of at least one MR imaging examination performed for measurement of lung volume, and for the FETO group, performed prior to the FETO procedure; (2) live birth. Exclusion criteria concerned foetuses in the FETO group where we failed to place the balloon into the trachea and where the FETO procedure was not repeated later on in gestation. Consequently, 32 foetuses were excluded because of termination of pregnancy (n = 21), failure to place the balloon into the trachea (n = 2), in utero foetal death (n = 2) or non-availability of an MR imaging examination (n = 7). This left 61 foetuses, 31 who underwent FETO and 30 who were conservatively managed. None of the foetuses included in this study have been previously reported.

MR imaging examination

Magnetic resonance imaging was performed with a clinical 1.5-T whole-body unit. A combination of six-channel phase-array body and spine coil was positioned over the lower pelvic area. The MR imaging protocol consisted of T2-weighted images. Geometric parameters of the T2-weighted images according to the machine used have been described previously [1618]. T2-weighted imaging was obtained using a single-shot half-Fourier TSE (HASTE) sequence in orthogonal transverse, coronal and sagittal planes according to the foetal orientation. No breath-hold was requested of the patient. In each centre, the radiologist adjusted the field of view and the number of sections and image orientation for each foetus as required for optimal measurement of lung volumes. Sequences that were degraded by foetal motion were repeated with the same parameters. Mean ± SD examination time was 20 ± 5 min.

MR imaging planimetry

The MR imaging blocks were collected from participating centres and centralised at the University Hospital Brugmann. Planimetric measurements of lung, liver and thoracic volumes were therefore all performed by one trained operator (M.M.C. with 11 years of experience in foetal MR imaging and planimetric measurements) who was not aware of the postnatal outcome. Lung, liver and thoracic volumes were calculated on the T2 HASTE sequences in the transverse plane using the sequences that allowed complete imaging of both lungs and liver without motion-induced artefacts.

The areas of the lung and liver were determined on each section by using free-form regions of interest (ROI) on PACS (Impax, Agfa-Gevaert, Mortsel, Belgium). The measured areas were added and multiplied by the slice thickness and intergap to determine the entire volume of the right and left lung, intrathoracic part of the liver and thorax. Volumes of right and left lung were added to obtain the TFLV of the corresponding foetus.

The degree of intrathoracic liver herniation was measured by volumetry using the xyphoid process and thoracic apex as a landmark, and the LiTR was calculated as previously described (Fig. 1) [16]. On the MR images, the left and right lungs, liver and thoracic volume could be measured in all foetuses. The time required to perform these measurements ranged between 15 and 20 min for the two lungs, the liver and the thoracic cavity.
https://static-content.springer.com/image/art%3A10.1007%2Fs00330-012-2709-6/MediaObjects/330_2012_2709_Fig1_HTML.gif
Fig. 1

T2-weighted single-shot TSE MR image (TR/TE = 1,000 ms / 88 ms) of a foetus with left-sided isolated diaphragmatic hernia at 24.9 weeks gestation (a) in a sagittal view showing the landmark at the xyphoid process (arrow) for the first plane of measurement (continuous line) and another two planes higher in the thorax (dashed and dotted lines). The same foetus in axial views at the level of the xyphoïd process (d) and two levels higher: (c) corresponding to the dashed line and (b) corresponding to the dotted line. Delineation of the thorax is performed with a dashed line and the liver with a continuous line

Data and statistical analysis

Each TFLV measurement was expressed as a percentage of the appropriate normal mean for gestational age (O/E ratio of TFLV × 100), which was previously established in a study of 215 foetuses with normal lung development at 21–38 weeks of gestation [17]. For foetuses who underwent FETO, when a second MR imaging examination was available at least 2 weeks after the FETO procedure, but before planned balloon removal, lung volume responsiveness to FETO was calculated. Lung volume responsiveness was defined with respect to pre-FETO O/E ratio of TFLV as follows: (TFLVpost−TFLVpre) × 100/TFLVpre, where TFLVpost is the O/E ratio of TFLV post-FETO but before the planned removal of the balloon and TFLVpre is pre-FETO O/E ratio of TFLV.

In foetuses who were managed conservatively in the prenatal period, univariate logistic regression analysis was used to investigate the effect on postnatal survival at discharge from the neonatal intensive care unit of the O/E ratio of TFLV (expressed as a percentage), LiTR (expressed as a percentage) and gestation at delivery (in weeks) as continuous numeric variables and side of CDH (left or right) and intrathoracic position of the liver (up or down) as categorical variables. Similarly, in foetuses who underwent FETO, univariate logistic regression analysis was used to investigate the effect on the postnatal survival of TFLVpre (expressed as a percentage), pre-FETO LiTR (expressed as a percentage), gestational age at FETO (in weeks), and duration of tracheal occlusion and gestation at delivery (in weeks) as continuous numeric variables, and side of CDH (left or right) as a categorical variable. For both groups, multiple logistic regression analysis was subsequently performed to determine the significant independent contribution of those variables that yielded a P value less than 0.05 in the univariate analysis. Subsequently, correlation between lung volume responsiveness after FETO and pre-FETO LiTR as a percentage was performed using linear regression analysis.

In all 61 foetuses with CDH, receiver-operating characteristic (ROC) curves were constructed and compared for the LiTR, the intrathoracic position of the liver and the O/E ratio of TFLV (all assessment before FETO for the FETO group) in the prediction of postnatal survival.

Data were analysed by using statistical software packages (SPSS 16.0, SPSS, Chicago, IL, USA; MedCalc 7.4, Mariakerke, Belgium; Excel for Windows 2007, Microsoft, Redmond, WA, USA). A two-sided P value of less than 0.05 was considered to indicate a statistically significant difference.

Results

Characteristics of the population

For the conservatively managed group, 28 had left-sided and 2 right-sided CDH. Median O/E ratio of TFLV was 36.4 % (range, 17.2–71.5 %), median LiTR was 2.7 % (range, 0–21.3 %) and median gestational age at delivery was 38.7 weeks (range, 33.1–41.1 weeks). Eighty percent of the foetuses (24 out of 30) were discharged alive from the neonatal intensive care unit. Conversely, neonatal death occurred in 6 at a median age of 5 days (range, 1–36 days), half before and half after surgical repair.

For the FETO group, 25 had left-sided and 6 right-sided CDH. Median O/E ratio of TFLV was 25.1 % (range, 5.9–56.9) and median LiTR was 18.0 % (range, 0.9–35.4). The median gestational age was 28.6 weeks (range, 25.3–30.7 weeks) at FETO, 33.6 weeks (range, 28.4–34.9 weeks) at removal of the balloon and 34.7 weeks (range, 28.4–38.9 weeks) at delivery. As a consequence, the median duration of FETO was 5.0 weeks (range, 0.7–8.9 weeks). Forty-two percent of the foetuses (13 out of 31) were discharged alive from the neonatal intensive care unit. Conversely, neonatal death occurred in 18 at a median age of 2 days (range, 0–29 days), 12 before and 6 after surgical repair.

Conservatively managed foetuses

Univariate regression analysis demonstrated that LiTR but not side of CDH, O/E ratio of TFLV, intrathoracic position of the liver or gestation at delivery was a significant predictor of postnatal survival (Table 1).
Table 1

Regression analysis for the prediction of survival in foetuses with conservatively managed isolated diaphragmatic hernia

Variable

n (%) or median (range)

Survival

Univariate analysis

OR (95 % CI)

P

Side of CDH

Left

28 (93.3)

4.600 (0.244–86.638)

0.308

Right

2 (6.7)

1

 

O/E ratio of TFLV (%)

36.4 (17.2–71.5)

1.572 (0.860–2.872)

0.142

LiTR (%)

2.7 (0.0–21.3)

0.845 (0.735–0.971)

0.017

Intrathoracic position of the liver

Up

17 (56.7)

1.135 E-9 (0.000–infinity)

0.999

Down

13 (43.3)

1

 

Gestational age at delivery (weeks)

38.7 (33.1–41.1)

1.447 (0.826–2.535)

0.197

CDH Congenital diaphragmatic hernia; TFLV total Fœtal lung volume, LiTR liver-to-thoracic volume ratio

Foetuses undergoing FETO

Univariate regression analysis demonstrated that pre-FETO LiTR and gestation at delivery but not side of CDH, TFLVpre, gestational age at FETO or duration of tracheal occlusion were significant predictors of postnatal survival. Logistic regression analysis demonstrated that LiTR and gestation at delivery provided independent prediction of postnatal survival (Table 2). In 19 foetuses undergoing FETO, a second MR imaging examination was available at a median of 5.0 weeks (range, 3.3–8.9 weeks) after the FETO procedure, but before planned balloon removal. There was a significant association between post-FETO lung volume responsiveness and pre-FETO LiTR (LVRpost = −5.9543 × pre-FETO LiTR + 177.12, where LVRpost is post-FETO lung volume responsiveness; r = −0.508, P = 0.026) (Fig. 2). In two cases, ultrasound follow-up caused suspicion of balloon deflation at around 32 weeks, which was confirmed at the time of planned balloon removal. For these two cases, LVRpost was −7.9 and −18.3 % and was underestimated from what the pre-FETO LiTR values (respectively 21.7 % and 14 %) would have predicted (Fig. 2).
Table 2

Regression analysis for prediction of survival in foetuses with isolated diaphragmatic hernia treated with fetoscopic endoluminal tracheal occlusion (FETO)

Variable

n (%) or median (range)

Survival

Univariate analysis

Multivariate analysis

OR (95 % CI)

P

OR (95 % CI)

P

Side of CDH

Left

25 (80.6)

4.615 (0.469–45.390)

0.190

  

Right

6 (19.4)

1

   

Pre-FETO O/E ratio of TFLV (%)

25.1 (5.9–56.9)

1.039 (0.961–1.123)

0.337

  

Pre-FETO LiTR (%)

18.0 (0.9–35.4)

0.792 (0.659–0.952)

0.013

0.810 (0.670–0.979)

0.029

Gestational age at FETO (weeks)

28.6 (25.3–30.7)

0.834 (0.473–1.468)

0.528

  

Tracheal occlusion duration (weeks)

5.0 (0.7–8.9)

1.542 (0.973–2.442)

0.065

  

Gestational age at delivery (weeks)

34.7 (28.4–38.9)

1.710 (1.098–2.664)

0.018

1.774 (1.003–3.135)

0.049

CDH Congenital diaphragmatic hernia; TFLV total Fœtal lung volume, LiTR liver-to-thoracic volume ratio

https://static-content.springer.com/image/art%3A10.1007%2Fs00330-012-2709-6/MediaObjects/330_2012_2709_Fig2_HTML.gif
Fig. 2

Plot of post-FETO lung volume responsiveness in respect to the pre-FETO value of the O/E ratio of total foetal lung volume (TFLV) as a percentage against the pre-FETO liver-to-thoracic volume ratio (LiTR) as a percentage in 19 foetuses with isolated congenital diaphragmatic hernia (CDH) treated by FETO. There was a significant inverse association between post-FETO lung volume responsiveness and pre-FETO LiTR. In two cases (diamonds), the balloon deflated at around 32 weeks, and their lung volume response was underestimated from what the pre-FETO LiTR values would have predicted

All foetuses with CDH

The area under the ROC curve for prediction of postnatal survival from LiTR was 0.868 (P < 0.001; standard error = 0.051) and from the intrathoracic position of the liver was 0.676 (P =0.998; standard error = 0.072). Pairwise comparison of ROC curves showed a significant difference between the areas under the ROC curves (difference = 0.192, P = 0.007; standard error = 0.071). The area under the ROC curve for the prediction of postnatal survival from the O/E ratio of TFLV was 0.783 (P =0.001; standard error = 0.058) (Fig. 3).
https://static-content.springer.com/image/art%3A10.1007%2Fs00330-012-2709-6/MediaObjects/330_2012_2709_Fig3_HTML.gif
Fig. 3

Receiver-operating characteristic curves for the prediction of postnatal survival according to cutoff values of LiTR (solid line), O/E ratios of TFLV before FETO (dashed line) and intrathoracic liver herniation (dotted line) in foetuses with isolated CDH showing that the best prediction of survival was provided by LiTR. Solid diagonal line = reference line

Discussion

This study has demonstrated that in isolated CDH foetuses, the LiTR provides useful prediction of postnatal survival, better than a categorical classification as intra-thoracic versus intra-abdominal liver. Further, it has demonstrated that in foetuses undergoing FETO, LiTR provided useful prediction of lung volume responsiveness to the balloon, assessed before balloon removal, hence of postnatal survival.

The strength of our study is that once we successfully placed the balloon, we did not further exclude any foetuses in the FETO group, even if delivery occurred before 32 weeks’ gestation. Instead, we included gestation at delivery in our logistic regression model of prediction of postnatal survival and showed that it was, independently from LiTR, predictive of postnatal survival. An earlier study has shown that significant prediction of premature delivery before 34 weeks was provided by the occurrence of premature rupture of the membranes, itself related to the duration of the FETO procedure [11]. The only two foetuses excluded from our analysis were those where we failed to place an intratracheal balloon, one of them because of placental bleeding and the second because of our inability to pass the vocal cords. Postnatally, this second case turned out to display Fryns syndrome. We felt we could exclude these two cases as our study was not an intention-to-treat study such as in a randomised controlled trial. Furthermore, in another two foetuses, post-FETO ultrasound follow-up led to suspicion of balloon deflation at around 32 weeks, nearly 2 weeks before planned removal. These two foetuses were also not excluded from our study. Interestingly, these foetuses showed smaller lung volume response from what was expected based on their LiTR measurement. It has been shown previously that balloon removal is accompanied by an immediate decrease in relative lung volume of nearly 50 % [15]. It is therefore not surprising to underestimate the O/E ratios of TFLV of these two foetuses, as evaluation was made after balloon deflation.

In CDH fetuses undergoing FETO, post-FETO lung volume responsiveness to the balloon has been shown to be a strong predictor of postnatal survival [15]. An earlier MR imaging study showed that there is an association between lung volume responsiveness at 1 month post-FETO and the gestational age at FETO. There was a cutoff at around 29 weeks’ gestation, above which lung response was minimal [15]. In the present study, gestation at FETO was between 25 and 30 weeks, thus ruling out the effect of gestational age at FETO. This way we could demonstrate that in those foetuses LiTR predicted lung volume responsiveness to FETO, which coincided with subsequent survival chances. If these preliminary findings are confirmed in larger studies, they may have important clinical consequences for the selection of foetuses for current foetal therapy protocols. In fact, our data show that the very severe CDH foetuses as predicted by a high LiTR are the ones that are less likely to respond positively to tracheal occlusion. These findings may therefore question the very optimistic results of the recent randomised controlled trial on FETO showing a nearly ten-fold increase in postnatal survival after FETO [10, 19]. It is therefore unlikely that FETO can improve the postnatal survival of these cases if properly selected for foetal surgery by nearly 50 %.

Other determinants of lung volume responsiveness could be found by exploring the technique used for tracheal occlusion. When the trachea was occluded by placement of hemoclips, impressive lung growth could be noted with an up to seven-fold increase in LHR at about 3 weeks post-FETO compared with the pre-FETO measurement [20]. In contrast, and in our study, when using an endoluminal balloon for tracheal occlusion, lung growth showed at maximum a 2.5-fold increase in the O/E ratio of TFLV for foetuses with an LiTR of 10 % or more. It is therefore likely that tracheal occlusion using hemoclips rather than an endoluminal balloon is more efficient in terms of occlusiveness of the trachea and subsequent lung response, but at the expense of higher tracheal side effects and thus postnatal mortality and morbidity [20].

Previous studies have shown that in foetuses with severe CDH treated with FETO, subsequent survival depends on the size of the lungs before FETO, whether this is measured by LHR using 2D ultrasound, O/E contralateral lung volume using 3D ultrasound or the O/E ratio of TFLV using MR imaging [15, 21, 22]. In the present study, the pre-FETO O/E ratio of TFLV was not predictive of subsequent survival. Such a finding could be explained by an underpowered study given the relatively low survival rate in our reported series or by the effect of another variable such as LiTR that would affect postnatal survival more than the pre-FETO O/E ratio of TFLV.

Finally, our study has shown that gestational age at delivery was predictive of postnatal survival in the FETO treated group while it was not in the conservative management group. This apparent contradiction can be explained by the fact that the proportion of foetuses delivered prematurely in the conservative management group was underrepresented as compared to the FETO group. In fact, while the range for gestational age at delivery in the conservative management group was between 33 and 41 weeks of gestation, this ranged between 28 and 39 weeks for the FETO-treated group.

We acknowledge that our study has some limitations. First, the profile of the population was biased with more severe cases being over-represented in foetal surgery centres. These severe cases were more likely to be undergoing FETO than being conservatively managed. Further studies are certainly needed in other centres that are not offering foetal therapy. Second, our study was retrospective and the MR imaging examinations were not performed for the purposes of measuring LiTR. On the other hand, the MR imaging examinations were prospectively performed for the purposes of measuring lung volumes and the acquisition mode is similar for lung volume or LiTR measurements. Furthermore, LiTR was previously shown to be highly reproducible even when retrospectively measured [16].

In conclusion, in foetuses with isolated CDH, quantification of intrathoracic liver as described in this study using LiTR seems to be a better predictive method than the method reporting on the liver as intra-thoracic versus intra-abdominal. For foetuses treated with FETO at mid-gestation, LiTR before balloon insertion is an indicator of lung volume responsiveness to FETO, hence postnatal survival. If these results are confirmed, they might have important clinical implications for the foetal treatment strategy in CDH.

Copyright information

© European Society of Radiology 2012