Prenatal diagnosis of esophageal obstruction is believed to improve the outcome for the affected newborn. However, the prenatal detection rate is only 10–40%, the diagnosis is usually not made before the third trimester and the false-positive rate has been high. This study investigated the prenatal detection rate and time of prenatal diagnosis at our center and its influence on outcome. In addition, incidence, detection rate and accuracy of the diagnosis in a large non-selected population were determined.
All cases diagnosed pre- or postnatally with esophageal obstruction and examined prenatally by ultrasound at the National Center for Fetal Medicine in Norway during 1987–2004 were evaluated.
Of 48 cases with esophageal obstruction, 21 (44%) were diagnosed prenatally (median, 32 + 0 weeks). All 21 had a small or empty stomach, 20/21 (95%) had polyhydramnios and 9/21 (43%) had a visible esophageal pouch. Associated anomalies were present in 38/48 cases (79%). The karyotype was abnormal in 11/48 cases (23%). Ten (21%) pregnancies with lethal fetal conditions were terminated. Two fetuses died in utero. Ten infants with associated anomalies died within 3 months after birth. The 26/48 (54%) survivors included 16/21 cases with a prenatal diagnosis of esophageal obstruction and 9/10 cases with isolated esophageal obstruction.
Esophageal atresia (the principal cause of esophageal obstruction) occurs in approximately 1/3500 births1–3. Associated anomalies are present in 40–65%2, 4–8 of cases. Prenatal diagnosis of esophageal atresia may improve the outcome by optimizing the pre- and postnatal care and treatment for the newborn9; thus, a high prenatal detection rate is desirable.
Polyhydramnios and a small or invisible fetal stomach are the most common ultrasound findings when esophageal atresia is suspected prenatally. However, these ultrasound findings have a low positive predictive value in diagnosing esophageal obstruction10 and the false-positive rate has been high. Approximately 80–90% of the cases with esophageal atresia have a tracheoesophageal fistula (TEF) to the distal esophageal segment6 and a blindly ending proximal part, referred to as the esophageal pouch. The proximal or upper pouch has long been known from postnatal diagnosis11–13 and was described prenatally in 198314. Methods for postnatal visualization of the distal pouch in cases with esophageal atresia without fistula are well established13, 15–17. In recent decades, prenatal visualization of the proximal esophageal pouch has received increasing interest18–22 because of its high specificity in the diagnosis of esophageal atresia19, 20. It has also been suggested that the level of the pouch may be of prognostic value23, 24. The accuracy of the prenatal diagnosis of esophageal atresia might be further improved by three-dimensional (3D) ultrasound25 and by MRI26.
It is possible to visualize the normal fetal esophagus during the first trimester27 and the esophagus can be detected in approximately 90% of normal fetuses in the second and third trimesters28, 29. Although prenatal diagnosis of esophageal atresia had already been described in the early 1980s11, 30, 31, the rate of prenatal detection has been reported to be low (9–42%)2, 7, 9, 10, 32 and the diagnosis is usually not made before the third trimester.
The aim of this study was to investigate the rate of prenatal diagnosis of esophageal obstruction at our center and the time when the diagnosis was made. We also wanted to discuss possible strategies for improving the prenatal detection rate. In addition, we wanted to determine the incidence, detection rate and accuracy of the prenatal diagnosis of esophageal obstruction in a large non-selected population.
Materials and Methods
The study included fetuses and newborns with esophageal obstruction who had been through at least one prenatal ultrasound examination at the National Center for Fetal Medicine (NCFM), Trondheim University Hospital, Trondheim, Norway, during January 1987–December 2004. NCFM is a tertiary referral center for prenatal diagnosis and therapy for the whole country. In Norway, women are offered one routine ultrasound examination at around 18 gestational weeks. Additional ultrasound examinations are only made on clinical indications. Complete obstetric history, ultrasound findings, autopsies and pre- and postnatal development are prospectively registered.
The study population represents a selected population, including referred cases, in addition to the cases from our non-selected population. The non-selected population was also investigated separately in order to determine incidence, general prenatal detection rate and accuracy of the diagnosis of esophageal obstruction. The non-selected population represents a geographically well-defined area, consisting of the city of Trondheim and eight surrounding municipalities. Between January 1987 and December 2004, 49 232 infants were born in this area. Within the non-selected population, approximately 97% of the pregnant women had a routine fetal examination at the NCFM and were later delivered and followed up at Trondheim University Hospital.
Gestational age was expressed in completed weeks and days, based on ultrasound measurements of the biparietal diameter at the routine fetal examination. Small for gestational age (SGA) was defined as a birth weight of less than mean − 2 SDs33. Statistical analyses were performed using Pearson's χ2 analysis.
The study population comprised 48 cases with esophageal obstruction, 46 with esophageal atresia, one with esophageal stenosis and one with a functional obstruction of the lower part of the esophagus as a consequence of a laryngotracheoesophageal cleft type IV (down to the carina)34, 35. Mean maternal age was 29 (range 17–40) years. There were 23 (48%) female and 24 (50%) male fetuses. In addition, one case had a male phenotype but a female karyotype. One fetus was a twin and one was a triplet. A tracheoesophageal fistula was present in 35/47 (75%) cases (not including the case with laryngotracheoesophageal cleft with complete communication between esophagus and trachea).
Prenatal diagnosis of esophageal obstruction
Prenatal diagnosis of esophageal obstruction was made in 21/48 (44%) fetuses at median 32 + 0 (range, 17 + 6 to 36 + 6) weeks. Polyhydramnios was present in 20/21 (95%) of these diagnosed cases; only one had a normal amount of amniotic fluid. In all 21 cases with prenatal diagnosis of esophageal obstruction, the fetal stomach was either small or not visible. In addition, an esophageal pouch was visualized in 9/21 (43%) cases (Figure 1, Table 1). Six (29%) of the 21 fetuses had isolated esophageal atresia and 15 (71%) had associated anomalies. The prenatal detection rate over time is shown in Table 1. Seventeen (81%) of the 21 cases with prenatally detected esophageal obstruction were diagnosed after 28 weeks, median 33 + 4 (range, 28 + 6 to 36 + 6) weeks, and all but one had clinical symptoms of polyhydramnios. In four fetuses, the diagnosis was made earlier, at 18 + 6, 22 + 2, 17 + 6 and 20 + 4 weeks, respectively. Of these four, two had associated anomalies that were detected at the routine fetal examination and both had abnormal karyotype (47 XX, + 18 and 48 XX, + 18, + fragment). In one fetus with trisomy 21, the stomach was not visualized, either at the routine fetal examination or later: this infant had a long gap atresia without fistula. The fourth fetus with an early diagnosis was part of an in vitro-fertilized, dichorionic, triamniotic triplet pregnancy. The first sign of abnormality in this case was an overly large stomach observed at 10 weeks' gestation36 and the suspicion of a high gastrointestinal obstruction was raised at that time. From 12 weeks and onwards, the stomach was empty and esophageal atresia was then suspected, but the diagnosis was made at 20 weeks' gestation when polyhydramnios occurred.
Table 1. Prenatal detection rate and survival of 48 cases with esophageal obstruction over time
Survival of prenatally detected
EP, esophageal pouch detected prenatally.
Prenatal diagnosis of various anomalies but not of esophageal obstruction
Twenty-two (46%) cases had a prenatal diagnosis of various anomalies at median 19 + 4 (range, 14 + 6 to 35 + 3) weeks, but the esophageal obstruction was undetected prenatally. Altogether 43/48 (90%) of the cases with esophageal obstruction had a prenatal diagnosis of esophageal obstruction and/or various associated anomalies. Among these, 25/43 (58%) developed polyhydramnios. Amniodrainage of mean 2800 (range, 500–4500) mL amniotic fluid was performed in 15 of these 25 cases. Eleven of the 43 cases (26%) had a normal amount of amniotic fluid. Seven cases (16%) had oligo- or anhydramnios, most (6/7) as a consequence of renal failure. None of these cases with oligo- or anhydramnios had a visible fetal stomach but esophageal obstruction was not diagnosed in any of them. In the subgroup of 17 fetuses without prenatal diagnosis of esophageal obstruction that continued the pregnancy, 5/17 (29%) had polyhydramnios, one had oligohydramnios and 11/17 (65%) had a normal or unknown amount of amniotic fluid.
No prenatal diagnosis
In 5/48 cases (10%) no structural anomalies were detected prenatally. All five had esophageal atresia with TEF; in four it was an isolated finding and in one there was an additional vertebral anomaly.
There was one prenatal false-positive case of esophageal obstruction. This case from 1990 was referred at 28 + 3 weeks with massive polyhydramnios. The fetal stomach was not visible. Five amniodrainages were performed, relieving 11.8 L altogether. A large mesoblastic nephroma was diagnosed postnatally; the infant had polyuria and the nephroma compressed the stomach.
Of all 48 cases, 10 (21%) had isolated esophageal atresia: six with TEF and four without TEF. Associated anomalies, including abnormal karyotype, were present in 38 cases (79%) (Table 2). Eleven (23%) fetuses had an abnormal karyotype. In addition, one case with esophageal stenosis had Di George syndrome. The most frequent associated anomalies were cardiac, followed by urogenital anomalies. Among the cases with normal karyotype and associated anomalies, the most common anomalies were urogenital, followed by imperforate anus (Table 3).
Table 2. The most common anomalies involved in 38 cases of esophageal obstruction with additional anomalies. Many cases had several anomalies
48,XX + 18 + fragment
46,XX, inv (12)(q15q24) + 18
Central nervous system
Table 3. The most common anomalies involved in 27 cases of esophageal obstruction with normal karyotype but with additional anomalies. Many cases had several anomalies
Central nervous system
The outcome is shown in Figure 2. Ten women (10/48; 21%) chose to terminate their pregnancies because of lethal associated fetal anomalies. In none of them was the esophageal obstruction diagnosed prenatally. Two fetuses (4.2%) died in utero. One with multiple associated anomalies died at 28 + 1 weeks, while the other fetus, which was growth-restricted and had isolated esophageal atresia with TEF, died at 22 + 1 weeks. Neither of these two cases had prenatal diagnosis of esophageal obstruction.
Thirty-six fetuses, including all 21 with a prenatal diagnosis of esophageal obstruction, were alive until delivery. Mean birth weight overall was 2300 (range, 775–3340) g and mean gestational age at delivery was 36 + 1 (range, 26 + 3 to 41 + 3) weeks. For survivors, the mean birth weight was 2444 (range, 775–3340) g and the mean gestational age at delivery was 36 + 1 (range, 27 + 0 to 41 + 3) weeks, and for those which suffered postnatal death, the mean birth weight was 1925 (range, 835–3010) g and the mean gestational age at delivery was 36 + 0 (range, 26 + 3 to 38 + 6) weeks (Figure 3). Thirteen infants (36%) were small for gestational age33. Four infants with abnormal karyotype died during or shortly after delivery. In addition, six infants with associated anomalies died within 3 months of birth.
Twenty-six infants (54%) survived. Of these, 16 had a prenatal diagnosis of esophageal obstruction, six had a prenatal diagnosis of various anomalies that did not include esophageal obstruction and four infants had no prenatally detected anomaly. The infant with a laryngotracheoesophageal cleft was successfully treated and survived.
Of the 21 cases with a prenatal diagnosis of esophageal obstruction, 16 (76%) survived compared to 10/27 (37%) of the cases without a prenatal diagnosis of esophageal obstruction (P = 0.14). When obvious lethal associated conditions were excluded, 16/16 (100%) of the cases with a prenatal diagnosis of esophageal obstruction survived and 10/14 (71%) of the cases without prenatal diagnosis of esophageal obstruction survived (P = 0.54).
Of the 10 cases with isolated esophageal obstruction, 9/10 (90%) survived compared to 17/38 (45%) of the cases with esophageal obstruction and associated anomalies (P = 0.20).
In the non-selected population, there were 14 cases with esophageal obstruction; this gives an incidence of approximately 1/3500 (14/49 232). Within this non-selected population, 6/14 cases (43%) were detected prenatally and there was no false-positive case. Two of 14 cases (14%) had abnormal karyotype. Associated anomalies, including abnormal karyotype, were present in 6/14 (43%) cases, while 8/14 (57%) cases had isolated esophageal obstruction. Eleven (79%) of the 14 cases survived. The median maternal age of the total non-selected population during 1987–2004 was 29 years.
In this study of esophageal obstruction, the overall prenatal detection rate was 44%. Of the prenatally detected cases, all had a small or non-visible stomach, 95% had polyhydramnios and 43% had a visible esophageal pouch. Even though more than half of the cases had only indirect signs of esophageal obstruction, there was only one false-positive case; this showed that, in contrast to other studies10, the ability to reveal other reasons for polyhydramnios and empty stomach was satisfactory.
Nevertheless, even though the detection rate was slightly higher in this study than in earlier reports2, 7, 9, 10, 32, the detection rate was still relatively low (44%) and the time of diagnosis was late, usually in the third trimester. Surprisingly, the detection rate in the non-selected population was the same (43%) as in the selected population (44%). Despite the possibilities of performing 3D-ultrasound or MRI in suspected cases, this was not done in any case and one might discuss whether a higher degree of suspicion leading to additional examinations might have improved the detection rate. On the other hand, the low rate of 1/49 (2%) false-positive cases was satisfactory. One may argue that the diagnosis of esophageal obstruction may be only suspected antenatally and not diagnosed until after delivery. We have characterized the cases as diagnosed when our findings have had clinical consequences. The clinical signs of polyhydramnios leading to additional targeted ultrasound examinations were the most important factor in making a correct prenatal diagnosis. The detection rate was relatively stable over time. The apparently poorer results in the last period of the study are probably due to the selected nature of the population, which included more complex cases than those referred in previous years.
Prenatal diagnosis of esophageal atresia has several advantages. An important one is being able to counsel and prepare the parents. The counseling should include information about the possible outcome, in terms of the morbidity and mortality of esophageal atresia, and the poorer outcome of populations diagnosed prenatally compared to neonatal populations should be addressed10. The possible association with unrecognized structural anomalies, such as imperforate anus and associated syndromes, should also be discussed. Esophageal atresia is found in connection with a number of associations/syndromes37, VATER38 or VACTERL being the most well known. The option to terminate the pregnancy must be addressed and may be considered the best choice by some parents. In the present study, 10 women (21%) chose to terminate their pregnancies because of associated conditions of a lethal nature. Major cardiac anomalies are frequent in cases with esophageal obstruction and are also known to affect neonatal survival39–41.
In this study, associated anomalies were present in 79% of the fetuses, which is higher than in former reports2, 4–7, 40, 42. Also, the 23% rate of abnormal karyotype in the present study was higher than that reported in other studies2, 5–7, 42. This may be explained by the relatively higher number of prenatally diagnosed cases included in our series than have been included in previously published series. Most previous studies are based on neonatal populations4–6, 40, 42. However, the higher rate of associated anomalies in our series may also be a consequence of the inclusion of referred cases with severe anomalies.
Esophageal atresia without fistula, usually with a long gap, is known to be associated with trisomy 2143, 44 and the only case with trisomy 21 in our study had a long gap atresia. Of the cases with esophageal obstruction in the present study, 12/48 (25%) also had imperforate anus, but only one of these 12 had abnormal karyotype (47 XX i(18p)i(18q)). These findings are supported by a large epidemiological study1 in which none of the cases with the combination of esophageal atresia and imperforate anus had an abnormal karyotype. This is noteworthy, since there is a well-known association with imperforate anus, usually without fistula, and trisomy 2145–47.
The rare case with laryngotracheoesophageal cleft could have been considered for exclusion from this study. However, this specific case had a functional obstruction at the entrance to the lower esophageal segment. Usually, these cases have clinical signs similar to those of esophageal obstruction, both pre- and postnatally48. Laryngotracheoesophageal clefts, Type I–III, are often associated with esophageal atresia with TEF, while Type IV, as in our case, could be considered as esophageal obstruction with extreme TEF34, 49.
Increased nuchal translucency at 11 to 13 + 6 gestational weeks has been shown in a case with isolated esophageal atresia50. In our study, one case at 10 weeks' gestation was already suspected to have a high gastrointestinal anomaly, due to a stomach too large for the gestational age. Overdistended stomach at 12 weeks was presented in a case with esophageal atresia without fistula but with additional duodenal stenosis51. In our case, the physiological herniation may have given rise to a similar situation, with a transient obstruction distal to the stomach. Normal values of the size of the fetal stomach have been described from 8 gestational weeks27 and, in our experience, the fetal esophagus is easily visualized at the 11 to 13 + 6-week scan27 (Figure 4). An early scan offered to all pregnant women may improve the prenatal detection rate, since nuchal translucency may be increased and thus trigger a more thorough evaluation; the size of the fetal stomach may be abnormal and the normal appearance of the fetal esophagus may not be visible. However, these aspects need to be studied and evaluated further. There may also be a potential for improving the prenatal detection rate of esophageal atresia if a third trimester scan were offered to all pregnant women. Most cases detected prenatally in this study had clinical signs of polyhydramnios, leading to a targeted examination. However, polyhydramnios was present in only 58% of the cases where the amount of amniotic fluid was known and in only 29% of ongoing pregnancies without prenatal diagnosis of esophageal obstruction. Targeted examination in the third trimester with evaluation of fetal swallowing and visualization of the fetal esophagus and possible esophageal pouch might improve the detection rate of esophageal obstruction, including cases without polyhydramnios, although there may be some limitations due to acoustic shadowing.
A low birth weight in infants with esophageal atresia is associated with a decreased survival rate39–41. One benefit of early prenatal diagnosis is the possibility to avoid premature delivery. A high proportion of infants with esophageal atresia are small for gestational age and associated anomalies are common. The presence of polyhydramnios may induce premature delivery. Adequate amniodrainage may prevent both premature delivery and early rupture of the membranes. In a situation in which it is not possible to prevent premature delivery, it is even more important to have a correct prenatal diagnosis and delivery at a place with a highly specialized neonatal intensive care unit, including specialists in pediatric anesthesiology40.
In all cases with esophageal obstruction, the obvious benefit of prenatal diagnosis is to prevent the risk of aspiration and to optimize the postnatal care of the newborn through delivery at a unit with highly specialized neonatal care and pediatric surgery facilities52. Some studies have not shown any benefit of a prenatal diagnosis of esophageal atresia2, 7, 10. However, the overall prenatal detection rate in previous studies has been low and the studies are mostly retrospective and only a few have included fetal cases. The cases detected prenatally have primarily been cases with serious associated anomalies. In this study, when obvious lethal associated anomalies were excluded, 100% of the cases with a prenatal diagnosis of esophageal obstruction survived, compared to 71% of the cases without a prenatal diagnosis of esophageal obstruction. The numbers were too small to draw conclusions and the difference was not statistically significant. Of all cases with a prenatal diagnosis of esophageal obstruction compared to those without a prenatal diagnosis, there might be a tendency (P = 0.14) towards a benefit for the cases detected prenatally, which should be further explored. It is important to keep in mind that the severity of the associated anomalies varies and that 90% of the cases with isolated esophageal obstruction survived, irrespective of whether or not the diagnosis was made prenatally.
The present study found the prognosis for cases with esophageal obstruction to be poor, with 46% mortality. In fetal populations it is important to identify fetuses with lethal conditions and to focus surveillance on fetuses that are expected to have a favorable outcome. Prenatal diagnosis of esophageal obstruction is important in this aspect and is desirable because of the clinical consequences for the newborn. In this study, there was a tendency to have improved outcome among cases that were detected prenatally. However, the diagnosis of esophageal obstruction has been relatively stable at a low level for a long time, so we need to focus on improving detection rate. In the first trimester, visualization of the fetal esophagus, and measurements of the size of the fetal stomach and nuchal translucency may be helpful to make a diagnosis. In the second and third trimesters, visualization of the fetal esophagus, swallow movements and possible esophageal pouch identification might contribute to a higher prenatal detection rate. Polyhydramnios is an important clinical sign but usually is not apparent until the third trimester. We believe that even without other obvious clinical manifestations, the prenatal detection rate could be improved by an increased awareness of the possibility of esophageal obstruction and a targeted search for the typical sonographic signs thereof.