Introduction

Among the malformations associated with esophageal atresia (EA) congenital heart disease (CHD) is only second after skeletal defects [1–3] and it is widely accepted that, together with birth weight, it largely conditions survival [4, 5].

Prognostic classification of EA proposed by Waterston in 1962 was based on birth weight, presence or absence of associated malformations and status of the lungs. This grouping system was used for years until it became obsolete because earlier diagnosis of the esophageal defect decreased the incidence of pneumonitis and improvements of neonatal care dramatically increased survival in all babies. In 1994, Spitz proposed a new prognostic classification more adapted to the current circumstances. It was based only on the influence on survival of birthweight below 1,500 g and the presence or absence of major congenital heart disease (MCHD). He divided EA patients into three groups: group I (weight above 1,500 without MCHD), group II (weight below 1,500 or MCHD) and group III (weight below 1,500 with MCHD). Mortality has been reported below 10% in group I, below 50% in group II and higher in group III [6–8].

Although accurate prenatal or early postnatal diagnosis of CHD in patients with EA should likely have an influence on the timing and strategy of EA repair or on the ensuing mortality, this issue has not been specifically investigated to this day.

The present study examines the impact of CHD diagnosis on the strategies of treatment and on the results of EA repair in a large group of patients.

Materials and methods

We studied the charts of patients treated for EA at our institution during the period 1982 to 2002 retrospectively, with particular attention paid to cardiologic diagnoses and mortality. We separately studied patients in whom prenatal ultrasonography had been carried out and those in whom CHD had been prenatally diagnosed by these means. We also separated patients in whom postnatal cardiologic assessment had been done prior to EA repair and those in whom this step was not taken. Among the patients with diagnosed CHD, we distinguished those with MCHD (those in which cyanosis and/or heart failure requiring surgical treatment were present) and those with non-threatening CHD. We also classified our patients according to the Spitz grouping system. CHD were classified into heart and great vessel malformations for descriptive purposes (Table 1). Obviously, both types of defects were often present in the same patient. Right aortic arch (RAA) and isolated dextrocardia were particularly considered because of their potential for interference with the standard surgical approach for EA repair. MCHD were separated from the non-threatening ones (Table 2).

Table 1 Heart and great vessel defects in 82/195 patients with EA
Full size table
Table 2 Major congenital heart disease in patients with EA
Full size table

The possible modifications of the strategy of surgical repair of EA in the different groups and for the specific cardiac defects were individually examined. Mortalities in all groups were compared by Chi square tests (χ2) with Yates’ correction when appropriate and a threshold of significance of P<0.05.

Results

One hundred and ninety-five children (99 boys, 50.5% and 96 girls, 49.5%) were included in the study; Out of these 195, 82 (42%) suffered CHD and, interestingly a significant majority (51 or 61%) were girls (χ2, P<0.05 vs 31 boys or 39%). One hundred and sixty-four (84%) had EA with distal tracheo-esophageal fistula (TEF), 19 (9.7%) had isolated EA, 8 (4.1%) had an H-type TEF, 2 (1%) had EA with proximal and distal TEF and 2 (1%) had EA with proximal TEF; incidence of CHD in each type of atresia was 71/164 (43%), 6/19 (31%), 3/8 (37%), 1/2 (50%), respectively, and there was no CHD in the last group. Differences in CHD incidence between isolated EA, H-type TEF and EA with distal TEF were not statistically significant (P>0.1).

Among the 82 children with CHD, 43 (52%) suffered MCHD (Table 2) and the remaining 39 had non-threatening CHD. In five babies diagnosed at birth with EA and CHD, esophageal repair was not attempted because they had multiple or very severe associated conditions (four trisomies and one Di George Syndrome) and died. Four out of these five children had MCHD and were excluded from the study because the possible influence of the cardiologic diagnosis on the treatment of EA could not be assessed in them. In consequence, only 78 children with CHD were considered for this purpose.

Only 32/78 pregnancies of children with CHD were ultrasonographically controlled and six of them (18.7%) had prenatal diagnosis of heart malformation that was confirmed after birth in all cases (3 VSD, 1 Fallot, 1 transposition of the great vessels and 1 coarctation of the aorta).

Thirty-eight of 78 babies (48.7%) including the 6 prenatally diagnosed had complete cardiologic assessment before EA repair and the CHD was appropriately identified in 26 (68%) while the diagnoses was incomplete in 12 (32%). Five of these 12 (41%) suffered MCHD (2 VSD, 1 Fallot, 1 complete atrioventricular canal and 1 truncus arteriosus). The pattern of CHD diagnosis changed over time. Between 1982 and 1992, 30/104 (28%) babies with EA had a diagnosis of CHD whereas 48/94 (51%) had such diagnoses between 1993 and 2002 (χ2 P<0.05).

Six children (3%) had an RAA and three of them were diagnosed prior to EA repair in three. In spite of this, right thoracotomy approach was preferred for EA with distal TEF repair except in one patient in whom hemodynamic instability due to MCHD precluded thoracotomy. This baby had only gastrostomy and died in the first week of life.

Overall, 15/113 babies with EA without CHD (13%), 4/39 with EA and CHD (10%) and 26/43 with EA and MCHD (60%) died. In this respect, differences in mortality in the groups without CHD and with threatening CHD were not statistically significant (P>0.1) whereas they were significant (P<0.05) between the group with MCHD and the other two groups.

Overall survival was 150/195 (77%) and when the patients were classified according to Spitz groups, mortality was 14/145 (9.5%) for group I, 26/44 (59%) for group II and 5/6 (83%) for group III (Table 3).

Table 3 Mortality according to Spitz risk classification
Full size table

Apparently, an accurate diagnosis of the CHD before EA repair had no significant influence in the results because among patients with an accurate diagnosis, mortality was 30.7% (8/26) and among those without preoperative diagnosis, it was 25% (3/12) (ns).

Discussion

Our series of EA is comparable to most other published in the literature in terms of incidence of CHD that has been reported to range between 13.2 and 41% [2, 9–17]. Our 42% is slightly higher perhaps because we included some malformations of the great vessels, like RAA or aberrant subclavian arteries (ARSA) due to their potential impact on the choice of the surgical approach for EA repair [9, 10]. The previously unreported higher incidence of CHD in girls with EA in comparison with boys came as a surprise. High proximal pouch in EA has been reported to be associated with aortic-arch, skeletal and azygos vein anomalies [11–13]. However, in our series, the incidence of CHD was similar in all EA types. The MCHD diagnosed in our series (Table 2) had a distribution similar to that in other series [15–20]. Our prevalence of MCHD (52%) was similar to that published by Spitz [4] and the distribution of the clinical material among the three groups and the mortality in each one was comparable as well (Table 3). The analysis of the nature of CHD in this series reveals a pattern consistent with the involvement of neural crest cell dysregulation in the pathogenesis [21, 22].

The relatively low rate of prenatal diagnosis of CHD is in part explained by the long period of time covered by the present study. Most diagnoses are concentrated in the most recent years. The relatively low rate on postnatal, preoperative diagnosis of some of the cardiovascular defects can be explained by the poor clinical expression of some of the defects with increased pulmonary flow during the first 3 or 4 days of life due to the high pressures in the lung vascular bed.

The increased diagnosis of CHD in the last years has also been noticed by other authors [7, 23] and it may reflect the increased use of ultrasonography as a routine diagnostic tool in the NICU as well as an increasing severity of the associated defects.

Six children (3%) had RAA, a proportion similar to that reported in other series (1.8–5%) [24–26]. Although it has been claimed that the finding of tracheal displacement at a certain level in plain X-ray of the thorax suffices the diagnosis of RAA [26, 27] other investigators disagree [28] and our experience goes in the same direction because no baby in our series was diagnosed on this basis. MRI [28–30] or aortography [28, 31] can accurately delineate the position of the aortic arch but they are not available everywhere around the clock and it is not realistic to pretend that that their routine use in newborns with EA is required. On the contrary, ultrasonography is the best non-invasive method of diagnosis of RAA [14] as attested by the high rate of preoperative diagnosis in our series (50%) in comparison with other series (8–15%) [24–26]. Nevertheless, since like other authors [24, 25] we used a right approach in all operated cases without any major technical problem related to the RAA, we do not believe that this diagnosis is determinant prior to EA repair. The same can be applied to our two cases of dextrocardia or ARSA.

As previously stated, only one 1,500 g baby had the treatment of the esophageal malformation delayed because of the presence of MCHD. His obvious hemodynamic instability precluded primary anastomosis and he died 10 days after gastrostomy under vigorous ventilatory and hemodynamic support. All other patients were managed by primary anastomosis and fistula ligation or feeding gastrostomy with continuous suctioning of the upper pouch depending on the malformation irrespective of the presence or not of CHD; the fact that mortality in 68% of patients with associated CHD diagnosed preoperatively was not different from that in those in whom the diagnosis was missed or wrong illustrates the relative importance of this diagnosis for repair and survival.

In summary, based on our experience we do believe that there is room for improving prenatal diagnosis of CHD. All patients with EA should be investigated ecographically prior to EA repair. With only a few exceptions, the diagnosis of CHD does not change the timing and the modalities of EA repair. Finally, the presence of an RAA, ARSA or dextrocardia does not change the chances of a successful esophageal anastomosis by the usual approach.