Introduction

The earliest description of a baby with esophageal atresia (EA) has been attributed to William Durston in 1670 [1]. However, it was not until 1939 that a baby survived following operative intervention. On successive days in November, 1939, two different babies in two different cities underwent an initial gastrostomy followed, at a later date, by esophageal substitution [2, 3]. The first successful primary repair came in 1941 and was performed by Dr. Cameron Haight in Ann Arbor through a left thoracotomy [4]. An anastomotic leak and stricture developed, and the baby stayed in the hospital for many months, but only required one esophageal dilation. She recovered uneventfully and led a normal life. Moreover, she had a son who was born with EA and a distal fistula.

EA is really a spectrum of anomalies and a number of classification systems (Gross, Vogt, etc.) have been described. However, it seems best to describe the anomaly rather than to classify it. This review will center on the thoracoscopic management of EA and distal tracheoesophageal fistula (TEF) which is by far the most common variant in the EA spectrum of diseases, and occurs in about 85–88% of all EA cases. The use of thoracoscopy for management of infants with isolated EA (without associated TEF) has been described in two recent series which can be reviewed [5, 6]. This paper is not meant to be an exhaustive review of EA/TEF, but more of a focus on the thoracoscopic repair of this condition.

The first thoracoscopic repair of EA was performed by Rothenberg and Lobe in Berlin in 1999 just prior to an IPEG meeting [7]. The first thoracoscopic repair of EA and TEF occurred in 2000 and was performed by Rothenberg in Denver a couple of weeks before the IPEG meeting that year [8]. In 2002, two series describing thoracoscopic repair of EA/TEF were published, one by van der Zee and Bax, and one by Rothenberg [9, 10].

In 2005, a multinational, multi-institutional, retrospective review was published describing the use of thoracoscopy for EA/TEF repair in 104 infants [11]. This remains the largest report describing thoracoscopic repair of EA/TEF in the literature. In this paper, eight very experienced minimally invasive surgeons from six institutions around the world collated their results to date. The mean patient age at operation was 1.2 days, the mean weight was 2.6 kg, the mean operative time was 130 min, the mean days of ventilation were 3.6, and the mean days of hospitalization were 18.1. Sixty-seven patients underwent ligation of the fistula using an endoscopic clip, whereas the fistula was sutured ligated in 37 patients. This group of patients was fairly representative of the spectrum of EA and its associated anomalies as 26 patients (25%) also required a laparoscopic fundoplication, seven underwent aortopexy (six performed thoracoscopically), four underwent laparoscopic duodenal atresia repair, and 10 patients had imperforate anus. Five patients underwent cardiac operations other than VSD/ASD repair. In this group, two patients developed a recurrent fistula at 3 months and 8 months postoperatively, and three patients died. A 7-month-old died of necrotizing enterocolitis, a 10-day-old died of congenital heart disease, and a 3-week-old died of esophageal disruption at the time of intubation in the intensive care unit.

The results from this series were very comparable to large series of open operations for EA/TEF that were published between 1985 and 2000 (Table 1) [12–16]. Approximately 30% of patients in all these series required at least one dilation, and the recurrent fistula rate of 1.9% in this thoracoscopic series was actually lower than any of the open reports. Similar numbers of patients underwent an aortopexy and fundoplication in these reviews.

Table 1 An attempt at comparing the current thoracoscopic series (Holcomb et al.) with previous reports using thoracotomy for patients undergoing repair of esophageal atresia with distal tracheoesophageal fistula (Gross Type C)
Full size table

Technique

The baby is positioned supine on the operating table and the oropharynx is suctioned. My preference to perform preoperative bronchoscopy will be discussed. Following bronchoscopy, the baby is intubated and placed in the left lateral decubitus position (assuming the aortic arch is on the left side). The position of the endotracheal tube (ETT) is not as important if high-frequency ventilation will be utilized, although the ETT should not be in the TEF or right main stem bronchus. If conventional ventilation will be employed, it is best to position the ETT distal to the origin of the fistula (if possible). This is more easily accomplished when the distal fistula enters the mid-trachea.

A small roll is gently placed under the baby’s shoulders and left flank to enhance the right intercostal spaces. A 12–14 French catheter is then placed through the baby’s mouth and into the upper esophageal pouch, and the table is tilted slightly to the left in reverse Trendelenburg. The surgeon stands on the left side of the table with the camera operator to his/her left. The scrub nurse is usually opposite the surgeon at the level of the camera operator. The monitor is situated opposite the surgeon for optimal viewing.

The first cannula is usually positioned at approximately the eighth intercostal space in the posterior axillary line. This 4–5 mm port is inserted using a cut-down technique, and a cannula with a blunt trocar is gently advanced into the thoracic cavity. This cannula will be used for the telescope and camera. Usually, only a small amount of CO2 insufflation is needed to decompress the lung if the high-frequency ventilator (HFV) is being employed. A 45° or 70°, 4–5 mm telescope is then introduced through the cannula and the thoracic cavity is visualized. At this point, the optimal location of the two working ports can be determined. 3-mm cannulas are introduced at the 4th and 5th or 4th and 6th intercostal spaces in the mid- to posterior axillary lines as working ports. If needed, a fourth cannula can be positioned posteriorly and inferiorly around the 10th or 11th intercostal space through which another instrument can be introduced for additional lung retraction, if needed.

The azygos vein is well seen, as is usually the proximal esophageal pouch with the silicone catheter in it, assuming that it is not high in the thoracic cavity. As the TEF has been identified preoperatively at bronchoscopy, the surgeon has a good idea of where the TEF enters the trachea. Often, if conventional ventilation is used, the TEF can be seen to distend with each ventilating breath. This is usually not as well seen with the HFV. The azygos vein is kept intact if possible. If not, it is divided with cautery. Once the fistula is identified, the pleura overlying the fistula is incised and the fistula is gently mobilized. It is important to be gentle in this mobilization to protect the vagus nerve fibers which are usually seen coursing along the trachea and along the distal esophagus. Once the TEF has been identified, it is my preference to next identify the proximal esophageal pouch to gain a better understanding of the amount of proximal pouch mobilization that will be needed in preparation for the esophageal anastomosis as opposed to proceeding with fistula ligation at this time. If the two esophageal segments are close to one another, little dissection of the proximal and distal esophageal segments will be needed. However, if the segments are widely separated, it may be necessary to dissect the proximal pouch high into the thoracic inlet to gain esophageal length for a tension-free anastomosis. Such dissection can be performed with cautery to gently mobilize the proximal pouch. It is important to be careful with this dissection to not injure the membranous trachea which is lying adjacent to the proximal esophagus. If the dissection is unusually difficult, a proximal TEF may be present. Once the proximal esophagus has been mobilized and is ready for anastomosis, the distal fistula is then ligated and divided. (It has not been ligated and divided earlier in the operation because the distal esophageal segment will often retract into the retropleural tissues during the dissection of the proximal pouch and can, on occasion, be difficult to find.)

The fistula can be divided in a number of ways. A non-locking metal clip is the most common way. As will be mentioned, a locking polymer clip, Hem-o-lok (Teleflex Medical Research, Triangle Park, NC), can also be utilized for this ligation. In addition, the fistula can be sharply divided and the trachea closed with one or two silk or Vicryl (Ethicon Inc., Somerville NJ) sutures. We have utilized all these techniques.

Once the fistula is divided, the esophageal anastomosis is performed in a similar fashion as with the open operation. Ideally, the posterior row of sutures is tied after all the posterior sutures have been placed. At times, it may be easier to tie each suture before placing the next one. We have utilized both intracorporeal and extracorporeal suture techniques, and prefer the intracorporeal approach, whereas others prefer the extracorporeal technique. Also, there does not appear to be an advantage for one suture versus another. My preference is to use 4–0 silk for the anastomosis, but others prefer 4 or 5–0 polydioxanone (PDS; Ethicon Inc., Somerville, NJ), or vicryl. Once the posterior row of sutures is tied, a 6 French tube is then inserted through the infant’s nares, across the esophageal anastomosis, and into the stomach. I find this useful for a variety of reasons including the ability to feed the baby through it several days following the operation if needed. The anterior portion of the anastomosis is then completed in an interrupted fashion.

After completing the anastomosis, attention is turned towards trying to separate the esophageal suture line from the fistula closure on the trachea. Usually, these two sites abut each other and an intact azygos vein often serves the purpose of naturally separating these two closure sites. Other surgeons prefer to interpose pleural or retropleural tissue to separate the two closure sites to prevent a recurrent TEF. A small silastic drain is then inserted through one of the incisions and positioned near the anastomosis. The 4–5 mm cannula site is closed with 4–0 and 5–0 absorbable sutures, and either 5–0 absorbable suture or steri-strips (3 M Company, St. Paul MN) are used to close the other small incisions. The anesthesia is then terminated and the baby is transported back to the intensive care unit, where the baby is gradually weaned from the ventilator over a period of hours to days, depending on the baby’s preoperative respiratory status.

Comments

Over the past 10 years, new information and concepts have become available about the thoracoscopic repair of EA/TEF. The following information represents my personal thoughts on this topic.

Is preoperative bronchoscopy helpful? In my opinion, it is, but not for the reasons typically espoused about looking for a proximal fistula, as a proximal fistula occurs in less than 5% of patients. If contemplating the thoracoscopic repair, preoperative bronchoscopy is helpful to give the surgeon an idea as to the expected gap length between the two esophageal segments. In most instances, the depth of the proximal pouch can be gauged from the chest radiograph with an indwelling oroesophageal tube. At bronchoscopy, if the distal fistula enters the carina, there is probably a large gap length between the two ends, especially if the upper pouch does not extend down into the mediastinum very far based on the chest radiograph. However, if the fistula enters the mid-trachea, there is probably a short gap length, which makes the thoracoscopic repair easier. This information can be helpful to surgeons early in their experience using thoracoscopy for EA/TEF repair. A secondary reason is that, at the time of bronchoscopy, vocal cord paralysis/paresis can be evaluated. There is new literature that indicates that vocal cord paralysis/paresis may occur more often than is usually appreciated [17]. In a few instances, it may have developed before the operation.

A second point centers on how best to ligate the distal TEF. Most commonly, a metal endoscopic clip is used from a commercially available clip applier. Some surgeons prefer to use a locking polymer clip which is thought to, perhaps, be more secure than the metal clip. Some surgeons prefer to place two ties around the TEF and divide the fistula distal to the second tie. Others prefer suture ligation of the fistula, and there are still others who prefer to divide the fistula at its entry into the trachea and close the tracheal side with sutures. There is no evidence that one technique is more advantageous than another. One concern with the metal clip is that it can erode through the fistula, and, perhaps, lead to a recurrent TEF, but Rothenberg has indicated that this has not happened in his experience (personal communication).

A third question is whether or not the type of suture used for the esophageal anastomosis makes any difference. In a report from our institution on 99 patients undergoing the open operation from 1985–2005, permanent suture was used in 62 patients, and absorbable suture was used in 32 patients. (A combination was used in five patients) [18]. There was no difference in weight at operation, gestational age, age at repair, or the mean number of associated anomalies between these two groups. In looking at the two groups, there was no difference in leak rate (p = 0.82), stricture formation (p = 0.47), or number of dilations needed per patient if a stricture formed (p = 0.21). Therefore, there appears to be no difference in leak rates or stricture formation based on the suture that is utilized for the esophageal anastomosis. A recent IPEG survey of its membership about current practice and technique found that 58% of respondents used 5–0 suture and 27% used 4–0 suture. Forty-four (44) % used absorbable monofilament and 39% used absorbable braided suture in this survey [19].

What are the advantages and disadvantages of the thoracoscopic approach? The thoracoscopic approach is transpleural and definitely requires a longer operative time for most pediatric surgeons. However, there is better visualization than with the open operation. The open operation is extrapleural in most instances (but can be transpleural), likely has a shorter operative time, and there is adequate visualization. If a long gap is found, this can be difficult to close with thoracoscopy, but it can also be difficult with the open operation. However, it seems that a long gap is easier to deal with via an open operation versus the thoracoscopic approach. Performing an anastomosis high in the chest may be easier with thoracoscopy, assuming that there is no significant tension on the anastomosis, but this can be quite difficult at thoracotomy. Finally, anesthesia is extremely important with thoracoscopy and is relatively standard with thoracotomy. In a recent review of the literature and meta-analysis of five retrospective comparative studies, there was no difference in complication rate, anastomotic leak, or anastomotic stricture between the thoracoscopic and open approaches [20].

The singular advantage of the thoracoscopic approach over the open operation is the reduction/absence of musculoskeletal morbidity following the thoracoscopic approach. A number of studies between 1980 and 2000 described significant musculoskeletal morbidity following a thoracotomy for EA/TEF [21–27]. In a more recent paper by Lawal et al., 62 infants and children were evaluated who had undergone either thoracoscopy or thoracotomy for several conditions between 2000 and 2006 [28]. The follow-up ranged from 1–7 years with a mean of 3.8 years. Statistically significant advantages in the group who underwent thoracoscopy included less chest asymmetry, less scoliosis, less nipple asymmetry, wider intercostal spaces, more favorable Manchester scar assessment, and more favorable patient satisfaction scores. The only variable that was not different between the two groups was the range of motion of the shoulder joints. Thus, this relatively recent paper confirms that there are musculoskeletal advantages of the thoracoscopic approach over thoracotomy when possible.

The two main disadvantages of thoracoscopy include the fact that it can be a very challenging technical operation and the concern about hypercapnia and acidosis. Not all surgeons can perform a thoracoscopic EA/TEF repair, and experience is required. The good results in the literature have come from surgeons and institutions with considerable experience with this disease as well as approach. It does not seem likely that one can become facile with this approach if one is performing a small number of thoracoscopic EA/TEF procedures every year or every few years.

Hypercapnia, acidosis and cerebral oxygenation are concerns for infants undergoing thoracoscopic operations. In a pilot, randomized control trial, 20 neonates were randomized to either open (five congenital diaphragmatic hernia (CDH), 5 EA/TEF) or thoracoscopic (five CDH, five EA/TEF) repair [29]. Arterial blood gasses were measured every 30 min intraoperatively and compared by multi-level modeling. For patients with CDH, the thoracoscopic approach was associated with a significant increase in intraoperative hypercapnia and severe acidosis. Interestingly, there was no significant difference in PaCO2, pH, or PaO2 in the patients undergoing open and thoracoscopic EA/TEF repair. In another study, mean blood pressure, FiO2, arterial oxygen saturation, and cerebral oxygen saturation were continuously monitored in 20 neonates undergoing thoracoscopic EA/TEF repair [30]. Although there was a decrease in arterial saturation and an increase in arterial PCO2 following intrathoracic CO2 insufflation, these changes did not result in significant fluctuations in cerebral oxygenation throughout the procedure. The authors concluded that insufflation of CO2 to 5 mmHg during thoracoscopy for EA/TEF seems to be safe in neonates since cerebral oxygenation was preserved during their procedures.

Because of concerns about hypercapnia and acidosis, and the fact that anesthesia can be difficult in these patients, we have used the HFV to facilitate stability during these thoracoscopic operations since 2007 [31]. With HFV, the patient should not develop hypercarbia. Also, the lung deflates nicely allowing excellent visualization, with the main problem being that the baby shakes from the oscillating ventilator. However, the surgeon becomes accustomed to this shaking and it is usually not a limiting factor. With the use of the HFV and proper positioning of the ET tube, the lung remains deflated and adequate operating space is achieved. The duration of the operation is not as important because the ventilator is able to remove the CO2 very efficiently and hypercarbia does not occur. Assuming a healthy 3–3.5 kg infant, initial HFV settings are a mean airway pressure of 13–15 cm of H2O, a ΔP of 30–35, and an FiO2 of 70%. The FiO2 can usually be weaned to 40% or so after 10–15 min. If the shaking of the baby becomes problematic, the Hertz can be decreased from its usual initial setting of 11–13 to 8–9 (1 Hz is defined as 60 breaths/s). A respiratory therapist accompanies the baby to the operating room and manages the ventilator during the operation. Therefore, for this reason, we rarely perform these operations on the weekend or holidays so that we have a full team available to help manage the baby.

Another part of the operation which may be underappreciated is separation of the esophageal anastomosis and the tracheal closure. Most surgeons try to insert a piece of mediastinal pleura between these two closures, and this is possible in many instances. Recently, the idea of keeping the azygos vein intact has developed in order for the intact vein to lie between these two closure sites. If the vein is intact and the esophageal anastomosis is performed lateral to the vein, then the vein naturally lies between the esophageal suture line and tracheal closure, and acts nicely to naturally separate these two areas. If all else fails, we have described placing small intestine submucosa (Surgisis, Cook Medical, Bloomington, IN) between the esophageal and tracheal suture lines to help prevent a recurrent TEF [32].

In a recent meta-analysis, the literature regarding the thoracoscopic repair of EA/TEF was reviewed [33]. There have been no prospective studies published to date and the best level of evidence for studies comparing the thoracoscopic versus open approach is level 3. The postoperative results for the series describing the thoracoscopic approach to EA/TEF are seen in Table 2.

Table 2 Series reporting on thoracoscopic EA/TEF repair
Full size table

Ure and colleagues recently reported their experience with 44 EA/TEF repairs over 10 years [34]. Using selective criteria (Table 3), 22 patients underwent a thoracoscopic repair and eight of these were converted to open. There were no leaks and there was one recurrent TEF. Seven of these 22 patients required at least one dilation. These criteria appear very reasonable and should serve as a guide for the use of the thoracoscopic approach for those who may not be as experienced with this technique as others. Also, when getting started with this approach, there appears to be an ideal case and not an ideal case. A baby less than 2.5 kg with a very high upper pouch and complex single ventricle physiology is not a good case to start one’s experience. A better case is a baby who is 3–3.5 kg and has no other anomalies with the esophageal segments close together as noted by chest radiograph and bronchoscopy. The suggestion is that the surgeon starts thoracoscopically and proceeds as far as he/she is comfortable.

Table 3 Selection criteria for thoracoscopic EA/TEF repair in routine pediatric surgery
Full size table

In summary, the thoracoscopic repair can be performed safely and effectively in about 75% of patients by experienced surgeons. Certainly, not every patient is a good candidate for this approach. The main advantage of thoracoscopy appears to lie in reducing the musculoskeletal sequelae seen following thoracotomy. As other authors have suggested, we will not be able to understand the differences and the advantages of each approach unless a prospective, randomized trial is performed.