R. STRESSIG*, R. FIMMERS, K. EISING*, U. GEMBRUCH* and T. KOHL

Transcription

1 Ultrasound Obstet Gynecol 2011; 37: Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: /uog.7747 Intrathoracic herniation of the liver ( liver-up ) is associated with predominant left heart hypoplasia in human fetuses with left diaphragmatic hernia R. STRESSIG*, R. FIMMERS, K. EISING*, U. GEMBRUCH* and T. KOHL *Department of Obstetrics and Prenatal Medicine, University Hospital of Bonn, Bonn, Germany; Department of Biostatistics, University Hospital of Bonn, Bonn, Germany; German Center for Fetal Surgery and Minimally Invasive Therapy, University of Giessen, Giessen, Germany KEYWORDS: diaphragmatic hernia; fetus; fetal echocardiography; fetal ultrasound; left heart hypoplasia ABSTRACT INTRODUCTION Objectives Left heart hypoplasia is commonly observed in fetuses with left diaphragmatic hernia. Because in this condition intrathoracic herniation of the liver serves as an important predictor for postnatal disease severity, we studied its potential association with left heart hypoplasia. Methods We prospectively assessed 32 fetuses with left diaphragmatic hernia between and weeks of gestation using echocardiography. The fetuses were divided into two groups: Group I exhibited an intrathoracic liver position ( liver-up ) and Group II an intra-abdominal liver position ( liver-down ). Cardiac inflow and outflow diameter ratios and cardiac Z-scores were compared between the two groups. Results Eleven of the 15 Group I (liver-up) fetuses, but only three of the 17 Group II (liver-down) fetuses with left diaphragmatic hernia exhibited predominant left heart hypoplasia with disproportionately smaller left than right heart dimensions (P = ). In addition, 14 of the 15 Group I fetuses, but only five of the 17 Group II fetuses exhibited preferential streaming of the ductus venosus towards the right heart (P = ). Conclusions In fetuses with left diaphragmatic hernia, intrathoracic liver herniation is commonly associated with predominant left heart hypoplasia, whereas an intraabdominal liver position is not. This observation may be explained by preferential ductus venosus streaming towards the right heart from elevation and leftward distortion of the normal course of the ductus venosus and inferior vena cava. Copyright 2011 ISUOG. Published by John Wiley & Sons, Ltd. Over the past decades, various mechanisms have been proposed in order to explain the common observation of smaller left than right heart structures in human fetuses with left diaphragmatic hernia. Among these mechanisms are reduced pulmonary flow to the left heart due to severe pulmonary hypoplasia, redistribution of fetal cardiac output, interference with pulmonary venous return by a distended posterior stomach, decreased expression of growth factors and compression of the left heart by herniated abdominal organs 1 5. Because in this condition liver herniation into the fetal chest serves as an important predictor of poorer postnatal outcome and, as a consequence, of fetal selection for prenatal intervention 6 8, the purpose of our study was to define a possible link between intrathoracic liver herniation and left heart hypoplasia. As such, we found that intrathoracic liver herniation is associated with preferential ductus venosus streaming towards the right atrium which might impact on left/right heart symmetry in this condition. PATIENTS AND METHODS Between January 2006 and March 2008, we assessed 32 human fetuses with isolated left diaphragmatic hernia between and weeks of gestation (Table 1), following maternal informed consent. Because the examinations were performed as part of the routine fetal echocardiographic ultrasound assessment in this condition at our center and the data were analyzed retrospectively, committee of human research approval was not required. The fetal echocardiographic studies Correspondence to: Dr R. Stressig, Department of Obstetrics and Prenatal Medicine, University Hospital of Bonn, Bonn, Germany ( R.Stressig@gmx.de) Accepted: 9 July 2010 Copyright 2011 ISUOG. Published by John Wiley & Sons, Ltd. ORIGINAL PAPER

2 Diaphragmatic hernia 273 Table 1 Cardiac inflow and outflow dimensions of 32 human fetuses with left diaphragmatic hernia with (Group I) and without (Group II) liver herniation into the chest GA (weeks) MV MV-z TV TV-z MV/TV Diff. (TV-z MV-z) AAo AAo-z MPA MPA-z AAo/ MPA Diff. (MPA-z AAo-z) CTA ratio DV to RA Group I: liver-up NP NP NP NP NA NA NA Group II: liver-down NP NP NP NP NA 0.21 Atrioventricular valve and great artery dimensions were expressed as gestational-age related Z-scores (MV-z, TV-z, AAo-z, MPA-z) based on data provided by Schneider et al. 9, employing a website calculator available at AAo, ascending aorta diameter; AAo-z, gestational age-related AAo Z-value; CTA, cardiothoracic area; Diff., difference; DV to RA, preferential streaming of ductus venosus towards right atrium; GA, gestational age; MPA, main pulmonary artery diameter; MPA-z, gestational age-related MPA Z-value; MV, mitral valve diameter; MV-z, gestational age-related MV Z-value; NA, no adequate measurement; NP, no prospective measurement; TV, tricuspid valve diameter; TV-z, gestational age-related TV Z-value. were performed using either an ATL-HDI 5000 system (ATL and Philips Medical Systems, Bothell, WA, USA) or a Sequoia 512 ultrasound system (Acuson-A-Siemens- Company, Mountain View, CA, USA). The fetuses were divided into two groups: Group I exhibited an intrathoracic liver position ( liver-up ) and Group II exhibited an intra-abdominal liver position ( liver-down ). We defined the intrathoracic position of the liver by demonstrating liver tissue and veins above the level of the diaphragm employing multimodal ultrasound. Measurements and statistics In the two groups, end-diastolic mitral and tricuspid valve dimensions, as well as early systolic ascending aorta and main pulmonary artery diameter, were measured in horizontal planes through the fetal chest. Atrioventricular valve and great artery dimensions were preferred over ventricular area measurements because they are less dependent on cardiac distortion and rightward positioning from the herniated abdominal organs. In addition, the cardiothoracic area ratio was measured 8. Atrioventricular valve and great artery dimensions were recorded and standardized using gestational agerelated Z-scores, based on data provided by Schneider and colleagues 9 employing a website calculator available at In addition, mitral-to-tricuspid valve diameter ratios (MV/TV) and ascending aorta-tomain pulmonary artery diameter ratios (AoA/MPA) were calculated in the two groups. Because in fetuses with diaphragmatic hernia, the increased intrathoracic pressure from herniated organs by way of preload reduction may result in symmetrical hypoplastic left and right heart dimensions (e.g. Group II; Cases 8 and 9), we arbitrarily

3 274 Stressig et al. coined and defined the term predominant left heart hypoplasia as a mitral valve Z-score < 2 in combination with a tricuspid valve Z-score that was at least 2 Z larger than the mitral valve Z-score in an individual fetus. In addition, in the two groups, spatial relationships of the ductus venosus inferior caval vein junction with the heart, as well as a preferential streaming direction of the ductus venosus towards the right atrium, were defined using 2D and color Doppler imaging in multiple planes. Following detection of this easily definable and characteristic flow deviation by one of the authors (T.K.) some years ago 10 12, assessment of ductus venosus streaming is routinely performed in our department. Data for the two groups are presented as mean and standard deviation (Table 2). Statistical analysis comparing fetal atrioventricular valve and great vessel Z-scores, diameter ratios and the absolute values of the differences between tricuspid valve and mitral valve, as well as main pulmonary artery and ascending aorta, Z- scores was performed using two-sided unpaired t-tests (Table 2). The absence or presence of preferential ductus venosus streaming towards the right atrium and the presence of predominant left heart hypoplasia were compared between the two groups using Fisher s exact test (twosided). Differences in cardiothoracic area ratio and gestational age at study between the two groups were compared using Mann Whitney Wilcoxon tests. RESULTS Eleven of 15 Group I (liver-up) fetuses, but only three of 17 Group II (liver-down) fetuses with left diaphragmatic hernia exhibited predominant left heart hypoplasia with disproportionately smaller left than right heart dimensions (P = ; Tables 1 and 2; Figure 1). MV/TV diameter ratio, AoA/MPA diameter ratio and ascending aorta and mitral valve Z-scores were significantly smaller in Group I than in Group II (P < , P = , P = ; Tables 1 and 2; Figure 2). By contrast, main pulmonary artery and tricuspid valve Z-scores were similar in both groups. Figure 1 Fetus with left diaphragmatic hernia, dextroposition of the heart and intrathoracic liver herniation (L) ( liver-up ) exhibiting disproportionately smaller left than right heart structures at weeks of gestation. (a) Cross-section through the fetal chest in the four-chamber plane exhibits a markedly smaller left ventricle (LV) than right (RV) ventricle. (b) Color Doppler imaging shows preferential streaming of the ductus venosus (DV) towards the right atrium (RA). A, anterior; I, inferior; L, left; LA, left atrium; P, posterior; R, right; S, superior; St, stomach. Inset schematic is meant to demonstrate the preferential streaming of the ductus venosus towards the right atrium characteristic of liver-up fetuses. Fourteen of the 15 Group I (liver-up) fetuses, but only five of the 17 Group II (liver-down) fetuses exhibited preferential streaming of the ductus venosus Table 2 Statistical calculations for selected cardiac inflow and outflow dimensions of 32 human fetuses with left diaphragmatic hernia with ( liver-up ) and without ( liver-down ) intrathoracic herniation of the liver Variable Group 1: liver-up Group 2: liver-down Difference P MV/TV ± (15) ± (16) ± < AAo/MPA ± (14) ± (15) ± MV-z ± (15) ± (17) ± < TV-z ± (15) ± (17) ± AAo-z ± (14) ± (17) ± MPA-z ± (14) ± (16) ± Data are presented as mean ± SD (n), with difference between groups expressed as mean ± SD. AAo/MPA, ascending aorta-to-main pulmonary artery diameter ratio; MV/TV, mitral valve-to-tricuspid valve diameter ratio. Atrioventricular valve and great artery dimensions were expressed as gestational age-related Z-scores (MV-z, TV-z, AAo-z, MPA-z) based on data provided by Schneider and colleagues 9 employing a website calculator available at

4 Diaphragmatic hernia 275 Table 3 Relationship between liver position and pathological ductus venosus streaming towards the right side of the heart in 32 fetuses with left diaphragmatic hernia with ( liver-up ) and without ( liver-down ) intrathoracic herniation of the liver Ductus venosus to right atrium streaming Group Present Absent Total Group 1: liver-up % 6.7% 46.9% 73.7% 7.7% Group 2: liver-down % 70.6% 53.1% 26.3% 92.3% Total % 40.6% Data given as n and percentages of both liver-up/liver-down group and present/absent ductus venosus streaming group. Two-sided test, P = Figure 2 Fetus with left diaphragmatic hernia and dextroposition of the heart without intrathoracic liver herniation ( liver-down ) and similarly sized left and right heart structures at weeks of gestation. (a) Cross-section through the fetal chest at the fourchamber plane exhibits similarly sized left (LV) and right (RV) ventricles. (b) Color Doppler imaging in adjacent planes shows normal infracardiac arrangement of ductus venosus (DV) and inferior vena cava (IVC) despite marked dextroposition of the heart. In most cases like this one with intra-abdominal liver position, preferential streaming of ductus venosus blood flow towards the right atrium (RA) was not observed. A, anterior; L, left; LA, left atrium; P, posterior; R, right; St, stomach. Inset schematic is meant to demonstrate the normal streaming direction of the ductus venosus towards the left atrium presumed characteristic of liver-down fetuses. towards the right heart (P = ; Figure 1 and Table 3). The cardiothoracic area ratio was similar in both groups (mean ± SD, Group I = ± 0.060, Group II = ± 0.019; P = ). Differences in the mean gestational age at the time of the study (29.3 weeks for liver-up and 30.3 weeks for liver-down fetuses) were not statistically significant. DISCUSSION Our study shows that intrathoracic herniation of the liver (liver-up), presumably by rerouting the majority of ductus venosus and inferior vena cava flow towards the right side of the heart, is in the majority of cases associated with predominant left heart hypoplasia in human fetuses with left diaphragmatic hernia. Although this abnormal course of the ductus venosus had already been noted in a previous postmortem study in a subgroup of fetuses with left diaphragmatic hernia and intrathoracic liver herniation, as well as in a clinical case, no inferences as to the possible hemodynamic consequences were made by the authors 13,14. Apart from the space-occupying effect of the herniated liver which directly competes with lung for space, the additional effect of abnormal hemodynamics on left heart size may help explain why in previous reports, a poorer prognosis has been stated for these fetuses compared with those without liver herniation (liver-down) 1,2,5,7. This detrimental effect of left heart hypoplasia in fetuses with diaphragmatic hernia has recently been challenged by Vogel et al. 15. Yet in their report, a large proportion of fetuses died before birth and in most of the neonates that died after delivery a postnatal echocardiogram had not been obtained. This exclusion of the severest cases may easily result in a selection bias that might falsify their hypothesis that hypoplasia of the left ventricle or other left heart structures in utero would not be associated with worse postnatal outcome. By contrast, in most fetuses with left diaphragmatic hernia without liver herniation (liver-down), preferential streaming of the ductus venosus towards the right heart occurs significantly less often. At the same time, the course of the infracardiac veins is less deviated than in the liver-up situation. These findings may explain why predominant underdevelopment of left-sided cardiovascular structures was observed less often, and both left and right heart dimensions remained similar in the Group II fetuses. Nevertheless, in some fetuses in this group, interference with cardiac filling from increased intrathoracic pressures resulted in the presence of both left and right heart hypoplasia, as demonstrated by calculating gestational age-related Z-scores (Table 1, Case 8). Of note, to our

5 276 Stressig et al. knowledge, the presence of right heart hypoplasia has not yet been described in this condition. This study had a number of limitations. Attempts at ventricular area measurements were abandoned because distortion of the heart due to displacement by the herniated abdominal organs did not allow reproducible imaging in four chamber planes in all patients. Yet in most Group I fetuses, the left ventricular area appeared at least a third smaller than the right ventricular area during imaging in adjacent horizontal planes through the fetal chest. Flow events across the foramen ovale and as such normal ductus venosus streaming across this flow region towards the left heart could not reliably be defined in many of the study fetuses due to several imaging restraints (advanced gestational age, unfavorable fetal lie, incidence angle). We, therefore, preferred to differentiate between the presence or absence of preferential ductus venosus streaming towards the right heart as this was clearly definable by color Doppler imaging at low Nyquist limits. In this context, the absence of abnormal ductus venosus streaming towards the right atrium was presumed to result in normal left-to-right shunting across the foramen ovale. A further limitation of this cross-sectional study is that the position of the liver is known to be mobile in fetuses with left diaphragmatic hernia. Accordingly, the spatial relationship between the ductus venosus and the heart might change over gestation and, hence, also the effect of ductus venosus streaming directions on fetal cardiac dimensions. In conclusion, in fetuses with left diaphragmatic hernia, intrathoracic liver herniation is commonly associated with predominant left heart hypoplasia, whereas an intraabdominal liver position is not. This observation may be explained by preferential ductus venosus streaming towards the right heart from elevation and leftward distortion of the normal course of the ductus venosus and inferior vena cava. REFERENCES 1. Allan LD, Irish MS, Glick PL. The fetal heart in diaphragmatic hernia. Clin Perinatol 1996; 23: Baumgart S, Paul JJ, Huhta JC, Katz AL, Paul KE, Spettell C, Spitzer AR. Cardiac malposition, redistribution of fetal cardiac output, and left heart hypoplasia reduce survival in neonates with congenital diaphragmatic hernia requiring extracorporeal membrane oxygenation. JPediatr1998; 133: Guarino N, Shima H, Puri P. The hypoplastic heart in congenital diaphragmatic hernia: reduced expression of basic fibroblast growth factor and platelet-derived growth factor. Pediatr Surg Int 2000; 16: Migliazza L, Xia H, Alvarez JI, Arnaiz A, Diez-Pardo JA, Alfonso LF, Tovar JA. Heart hypoplasia in experimental congenital diaphragmatic hernia. J Pediatr Surg 1999; 34: Siebert JR, Haas JE, Beckwith JB. Left ventricular hypoplasia in congenital diaphragmatic hernia. JPediatrSurg1984; 19: Bootstaylor BS, Filly RA, Harrison MR, Adzick NS. Prenatal sonographic predictors of liver herniation in congenital diaphragmatic hernia. J Ultrasound Med 1995; 14: Hedrick HL, Danzer E, Merchant A, Bebbington MW, Zhao H, Flake AW, Johnson MR, Liechty KW, Howell LJ, Wilson RD, Adzick NS. Liver position and lung-to-head ratio for prediction of extracorporeal membrane oxygenation and survival in isolated congenital diaphragmatic hernia. Am J Obstet Gynecol 2007; 197: 422.e1 422.e4. 8. Awadh AM, Prefumo F, Bland JM, Carvalho JS. Assessment of the intraobserver variability in the measurement of fetal cardiothoracic ratio using ellipse and diameter methods. Ultrasound Obstet Gynecol 2006; 28: Schneider C, McCrindle BW, Carvalho JS, Hornberger LK, McCarthy KP, Daubeney PE. Development of Z-scores from fetal cardiac dimensions from echocardiography. Ultrasound Obstet Gynecol 2005; 26: Kohl T, Franz A, Stressig R, Tchatcheva K, Schaible T, Geipel A, Gembruch U. Abnormal ductus venosus streaming toward the right heart may provide an anatomical and pathophysiological rationale for the prognostic differentiation between liver-up and liver-down left diaphragmatic hernias in human fetuses. Ultraschall Med 2008; 29: Kohl T, Franz A, Geipel A, Berg C, Schaible T, Gembruch U, Stressig R. Abnormal ductus venosus streaming toward the right heart is associated with low pulmonary blood flow and vascular reactivity anomalies in human fetuses with left diaphragmatic hernia. Ultraschall Med 2008; 29: Kohl T, Franz A, Geipel A, Stressig R, Berg C, Tchatcheva K, Gembruch U. Abnormal ductus venosus streaming toward the right heart offers a new and plausible explanation for left heart hypoplasia in human fetuses with left diaphragmatic hernia. Ultraschall Med 2008; 29: Sherer DM, Eglinton GS, Gonçalves LF, Lewis KM, Queenan JT. Prenatal color and pulsed Doppler sonographic documentation of intrathoracic umbilical vein and ductus venosus, confirming extensive hepatic herniation in left congenital diaphragmatic hernia. Am J Perinatol 1996; 13: Beaudoin S, Bargy F, Mahieu D, Barbet P. Anatomic study of the umbilical vein and ductus venosus in human fetuses: ultrasound application in prenatal examination of left congenital diaphragmatic hernia. Surg Radiol Anat 1998; 20: Vogel M, McElhinney DB, Marcus E, Morash D, Jennings RW, Tworetzky W. Significance and outcome of left heart hypoplasia in fetal congenital diaphragmatic hernia. Ultrasound Obstet Gynecol 2010; 35: