The survival of infants born with esophageal atresia. Esophageal Atresia and Tracheoesophageal Fistula: Surgical Experience Over Two Decades

Transcription

1 Esophageal Atresia and Tracheoesophageal Fistula: Surgical Experience Over Two Decades Josephine Y. Tsai, MD, Leah Berkery, BS, David E. Wesson, MD, S. Frank Redo, MD, and Nitsana A. Spigland, MD Department of Surgery, The New York Hospital-Cornell Medical Center, New York, New York Background. Despite improvements in survival, for infants born with esophageal atresia tracheoesophageal fistula, or both, the morbidity associated with repair of these anomalies remains high. Methods. This report retrospectively analyzes 81 patients with esophageal atresia, tracheoesophageal fistula, or both presenting to our institution between 1975 and 1995, with a focus on anastomotic complications. Results. There were 46 male and 35 female patients with a mean gestational age of 37 weeks and mean birth weight of 2443 g. Forty-four patients underwent primary esophageal anastomoses, 7 underwent delayed primary anastomoses, 12 patients underwent staged repairs, and 5 underwent repair of H-type fistulas. Among 62 patients with anastomoses, complications included stricture in 25/62 patients (40%), leakage in 12/62 patients (19%), and recurrent tracheoesophageal fistulas in 6/62 patients (10%). Stricture rates for esophagocolonic anastomoses versus esophagoesophageal anastomoses were 4/8 cases (50%) versus 21/54 cases (39%). This difference was not statistically significant. All esophagoesophageal strictures were managed successfully with dilations; three of four esophagocolonic strictures required anastomotic revision. The leakage rate for esophagocolonic anastomoses versus esophagoesophageal anastomoses was 6/8 cases (75%) versus 6/54 cases (11%). This difference was statistically significant (p ). Two patients required revision of their colon grafts secondary to necrosis. Eighteen of 81 patients (22%) died. Operative mortality was 9/74 (12%). Causes of death included associated anomalies (n 15), recurrent aspiration and sepsis secondary to missed fistula (n 1), and unknown (n 2). Conclusions. Although the morbidity associated with surgical repair of these anomalies is high, this does not affect the overall survival. The high complication rate associated with colonic interposition suggests that one should preserve the native esophagus as a primary conduit whenever feasible. (Ann Thorac Surg 1997;64:778 84) 1997 by The Society of Thoracic Surgeons The survival of infants born with esophageal atresia (EA), tracheoesophageal fistula (TEF), or both has improved dramatically since Cameron Haight s first successful repair in 1941 [1]. Improvements in survival are largely attributable to refinements in neonatal intensive care, anesthetic management, ventilatory support, and surgical techniques. Survival may now be achieved in infants with low birth weight [2], with mortality limited to those patients who have severe life-threatening anomalies. Despite improvements in survival, the morbidity associated with surgical repair of these anomalies remains high [3]. In this review, we update our previously reported experience with EA and TEF [4]. We retrospectively analyzed the charts of 81 patients with EA, TEF, or both who presented to our institution over the past two decades, with an emphasis on anastomotic complications so that future determinations could be made to minimize morbidity and mortality. Presented at the Poster Session of the Thirty-third Annual Meeting of The Society of Thoracic Surgeons, San Diego, CA, Feb 3 5, Address reprint requests to Dr Spigland, Division of Pediatric Surgery, The New York Hospital-Cornell Medical Center, 525 E 68th St, F763, New York, NY ( naspigla@mail.med.cornell.edu). Material and Methods Between 1975 and 1995, 81 patients with EA, TEF, or both were treated at the New York Hospital-Cornell Medical Center. The data were collected retrospectively from hospital charts. The information obtained was compared with data from a previous review from our institution published in 1975 [4]. Fisher s exact test was used for comparative statistical analysis with a significance level of p less than In our current study there were 46 male patients (57%) and 35 female patients (43%). The mean gestational age was 37 weeks (range, 28 to 42 weeks). The mean birth weight was 2,443 g (range, 915 to 4,035 g). The mean Apgar score at 1 minute was 7, and the mean at 5 minutes was 8. All cases were classified into anatomic anomaly type based on the Gross classification [5] (type A-D, plus H-type TEF). The most common variant was EA with a distal TEF (type C), which occurred in 67 patients (83%). Pure EA (type A) was present in 7 patients (9%). Five patients had an H-type TEF without EA (6%). One infant (1%) had EA with proximal TEF (type B), and 1 infant (1%) had both a proximal and distal TEF (type D). Infants were also assigned to risk groups A, B, or C as described by Waterston and associates [6]: 1997 by The Society of Thoracic Surgeons /97/$17.00 Published by Elsevier Science Inc PII S (97)

2 Ann Thorac Surg TSAI ET AL 1997;64: ESOPHAGEAL ATRESIA AND TRACHEOESOPHAGEAL FISTULA 779 Group A 1. Birth weight 2,500 g and well Group B 1. Birth weight 1,800 to 2,500 g and well, or 2. Birth weight 2,500 g but moderate pneumonia and other congenital anomaly Group C 1. Birth weight 1,800 g or 2. Birth weight 1,800 g with severe anomaly or pneumonia Twenty-four patients (30%) were classified in group A, 31 patients (38%) in group B, and 26 patients (32%) in group C. Forty of 81 patients (49%) had associated congenital anomalies. Congenital cardiac anomalies were the most frequent, occurring in 24 (30%) of 81 patients. These included atrial septal defects (ASDs), ventricular septal defects (VSDs), patent ductus arteriosus, patent foramen ovale, dextrocardia, pulmonic stenosis, subaortic stenosis, bicuspid aortic valve, hypoplastic left ventricle, and tricuspid atresia. Gastrointestinal anomalies occurred in 14 of 81 patients (17%), and included imperforate anus, duodenal atresia, pyloric atresia, biliary atresia, and annular pancreas. Major renal anomalies occurred in 9 of 81 patients (11%) and included hydronephrosis, renal agenesis, double collecting system, and horseshoe kidney. Minor skeletal deformities occurred in 8 of 81 patients (10%) and included missing radius, polydactyly, and 13th rib. Chromosomal anomalies occurred in 7 (9%) of 81 patients: trisomy 18 in 4 patients (5%), trisomy 21 in 2 patients (2%), and fragile X syndrome in 1 patient (1%). Severe life-threatening anomalies occurred in 8 of 81 patients (10%): bilateral renal agenesis and hypoplastic lung in 1 patient (1%), trisomy 18 in 4 patients (5%), dextrocardia, ASD, pulmonic stenosis, and left renal agenesis in 1 patient (1%); dextrocardia, ASD, VSD, imperforate anus, hypoplastic left lung, and agenesis of right lung in 1 patient (1%); and hypoplastic left heart and ASD in 1 patient (1%). Among 81 patients in our series, 7 died before surgical intervention, of severe associated congenital anomalies. Four patients died of severe anomalies during staging procedures, and 2 patients were transferred to other institutions after preliminary gastrostomy or esophagostomy. Of the remaining 68 patients, 44 patients underwent primary esophageal anastomoses, 7 patients underwent delayed primary anastomoses, and 5 patients underwent repair of H-type fistulas. Twelve patients underwent staged repairs for long-gap type A and C anomalies including esophageal anastomosis after bougienage (2 patients), colon interposition (8 patients), and staged esophageal lengthening (2 patients; 1 with anastomosis). Primary repair included fistula division and end-toend esophagoesophageal anastomoses within the first 48 hours. Delayed primary repairs included fistula division and esophageal anastomoses after 48 hours (median time, 9 days) after an initial gastrostomy. These were performed in patients with severe pneumonia or other anomalies that prohibited an immediate definitive operation. Staged repairs included initial thoracotomy and fistula division with gastrostomy or esophagostomy, and delayed esophageal anastomosis or colon interposition. In most cases, the position of the aortic arch was confirmed by cardiac echocardiography preoperatively and thoracotomy was performed opposite the side of the aortic arch. Primary repairs were performed by a retropleural approach whenever feasible. A single layer endto-end esophageal anastomosis was constructed using interrupted 5-0 silk sutures. Staged repairs were performed in 12 cases of wide-gap type C or type A anomalies (gap length greater than two vertebral bodies or greater than 3 cm). The mean gap length for the wide-gap anomalies in our series was four vertebral bodies. After initial thoracotomy with fistula division or gastrostomy, bougienage lengthening of the upper esophageal pouch was begun daily. Despite continuous sump drainage of the upper pouch with bougienage lengthening, 10 patients underwent cervical esophagostomies because of repeated bouts of aspiration. These patients subsequently underwent colon interposition or primary esophageal anastomoses after multistaged extrathoracic esophageal lengthening. Colon interpositions usually involved the right colon including the ileocecal valve in an isoperistaltic, substernal fashion. Follow-up of patients in our series ranged from 2 weeks to 9 years. Results Anastomotic strictures occurred in 25 of 62 patients (40%) with esophagoesophageal (EE) or esophagocolonic (EC) anastomoses. A stricture was defined as an anastomotic narrowing on an upper gastrointestinal (UGI) series that required one or more dilations. The median number of postoperative dilations performed in our series was two with a range of one to nine dilations. Nine of 25 patients (15%) required more than two dilations. Nine of 25 patients (15%) with strictures had associated anastomotic leaks. Anastomotic leaks were present in 12 of 62 patients (19%), and subsequent strictures that required dilation developed in 9 of 12 patients (75%) with anastomotic leaks. Five of 25 patients with strictures (20%) had associated gastroesophageal reflux (GER). The stricture rate in the staged group was 5/11 patients (45%) versus 20/51 patients (39%) for the primary and delayed primary groups. These differences were not statistically significant. The stricture rates for EC anastomoses versus EE anastomoses were 4/8 cases (50%) versus 21/54 cases (39%). These differences were also not statistically significant. All EE strictures were managed successfully with dilations; 3/4 EC strictures required anastomotic revision. Anastomotic leaks were identified clinically by the presence of saliva in the chest tube, and confirmed by UGI series. The leak rate in the staged repair group (wide-gap type C and type A) was 6/11 patients (54%) versus 6/51 patients (11%) in the primary and delayed primary group. These differences were statistically significant (p 0.004). The leakage rates for EC anastomoses versus EE anastomoses were 6/8 cases (75%) versus 6/54 cases (11%). These differences were also statistically sig-

3 780 TSAI ET AL Ann Thorac Surg ESOPHAGEAL ATRESIA AND TRACHEOESOPHAGEAL FISTULA 1997;64: Fig 1. Comparison of leak and stricture rates in the current series ( ) as compared with the previously reported series [4] ( ). These differences were not statistically significant. nificant (p ). Among the 12 patients with anastomotic leaks, 3 of the 6 with EE leaks (50%) developed strictures, which were successfully managed by dilation. Five of the six EC leaks (83%) eventually required surgical revision for intractable stricture or graft necrosis. A comparison of the leak and stricture rates between the previous review from our institution and the current series is seen in Figure 1. The differences were not statistically significant. Among 8 patients who underwent colon interposition, strictures developed in 4 of 8 patients (50%) anastomotic leaks developed in 6 of 8 patients (75%), and 5 of 8 patients (62%) required reoperation. Indications for reoperation included nondilatable strictures in 3 of 8 patients (37%) and colon graft necrosis in 2 of 8 patients (25%). The 2 patients in whom colon graft necrosis developed underwent secondary surgical procedures using the left colon and transverse colon, respectively. Recurrent tracheoesophageal fistulas were diagnosed in 6 patients at 10 days, 1 month, 1 month, 3 months, 1 year, and 4 years postoperatively by UGI series and confirmed by bronchoscopy. Three of these patients were treated by reoperation, which involved fistula division with interposition of a pleural flap between the suture lines. In 2 patients the follow-up management is unknown. One patient was managed nonoperatively for a pinpoint fistula noted on bronchoscopy. Two patients had missed fistulas (2.5%). One of these patients was presumed to have a type A anomaly, but later was discovered to have a type B anomaly. His postoperative course was complicated by pneumonia, and a missed fistula was discovered on UGI series. He underwent successful reoperation and division of his proximal TEF on postoperative day 7. In the second patient, recurrent episodes of pneumonia developed. A UGI series performed 1 month postoperatively did not reveal the presence of fistulas. The patient subsequently died 6 months postoperatively of recurrent bronchopulmonary pneumonia and sepsis. Two missed fistulas were noted on autopsy. Gastroesophageal reflux was documented by UGI series or ph probe in 9 patients. Gastroesophageal reflux was diagnosed between 1 month and 6 years postoperatively. Six of these patients were treated medically. Three of the 9 patients required antireflux procedures. Tracheomalacia was diagnosed in 9 patients (11%) by bronchoscopy between 1 month and 3 years postoperatively. Two patients manifested dying spells requiring a surgical procedure (aortic suspension in 1, stent placement in 1). The remaining 7 patients had less severe symptoms and were managed nonoperatively. Eighteen of 81 patients (22%) died. Seven patients died before any surgical intervention of known severe associated anomalies (trisomy 18 in 1; dextrocardia, ASD, VSD, hypoplastic lungs, and imperforate anus in 1; dextrocardia, ASD, pulmonic stenosis, and renal agenesis in 1; subaortic stenosis in 1; and bilateral renal agenesis and hypoplastic lungs in 1) and unknown causes (n 2). Four patients died after preliminary staging procedures (eg, gastrostomy, thoracotomy with fistula ligation) before definitive surgical repair. Causes of death included trisomy 18 in 2, VSD and bicuspid aortic valve in 1, and prematurity and intracranial hemorrhage in 1. Seven patients died after definitive surgical repair. The causes of death among these patients included recurrent aspiration and sepsis secondary to missed fistula in 1, trisomy 18 in 1, biliary atresia and liver failure in 1, hyaline membrane disease and bronchopulmonary dysplasia in 1, tricuspid atresia, VSD, and rudimentary right ventricle in 1; ASD, anamolous pulmonary venous return, and hypoplastic left heart in 1; and prematurity and subarachnoid hemorrhage in 1. The operative mortality was 9 of 74 patients (12%) who underwent surgical procedures. The late mortality was 2 of 74 patients (3%). One of these patients died at 1 year of age of liver failure secondary to biliary atresia. One patient with multiple cardiac anomalies died of an unknown cause at 2 years of age. If one considers only those infants who underwent definitive correction of their anomalies, overall mortality is 7 of 67 patients (10%) with an operative mortality of 5 of 67 patients (7%) and a late mortality of 2 of 67 patients (3%). A comparison of the total mortality, operative mortality, late mortality, and mortality after definitive repair between the two series is seen in Figure 2. The survival based on Waterston s criteria was 100% for group A, 89% for group B, and 11% for group C. Fig 2. Comparison of overall mortality, operative mortality, late mortality, and mortality after definitive (def.) repair in the current series ( ) as compared with the previously reported series [4] ( ). These differences were not statistically significant.

4 Ann Thorac Surg TSAI ET AL 1997;64: ESOPHAGEAL ATRESIA AND TRACHEOESOPHAGEAL FISTULA 781 Comment An analysis of the surgical experience with EA and TEF at the New York Hospital-Cornell Medical Center over the past two decades, as compared with the report from our institution published in 1975 [4], shows that there were no statistically significant differences with respect to sex predominance, birth weight, proportion of infants with low birth weight, proportion of infants with major lifethreatening congenital anomalies, anomaly type, or type of operative repair. The respective anastomotic leak and stricture rates among both series were comparable. A higher leak rate (19% versus 10%) is noted in the current series, but this difference is not statistically significant. The overall mortality, operative mortality, and mortality rates after definitive repair are also comparable between both groups. The overall survival rates were 75% and 78%, respectively, and survivals for those infants undergoing definitive correction of their anomalies were 93% and 90%, respectively. Despite advances in neonatal intensive care as well as surgical and anesthetic techniques, a significant proportion of infants in both series died because of associated life-threatening congenital anomalies. Current trends toward early detection of complex genetic and chromosomal malformations in utero, early total correction of cardiac defects, and improved neonatal intensive care management of premature infants may help increase survival rates for patients in the Waterston C risk group. If we exclude those infants who died preoperatively or awaiting definitive repair after preliminary staging procedures, survival rates that exceed 90% may be achieved, which are comparable with those published among several large series [3, 7 9]. Despite improvements in survival over the past five decades ranging from 36% in the pre-1950 era [7] to as high as 95% [3] in 1995, the incidence of anastomotic complications remains consistently high, with leak rates varying from 8.5% to 36% [3, 7 15], stricture rates varying from 8% to 37% [3, 8 14, 16], and the rate of recurrent TEF varying from 3% to 12% [3, 7, 8, 11, 13], depending on the criteria used to define these complications. The use of silk sutures [10, 15], excessive anastomotic tension [8, 14], and excessive distal esophageal mobilization [17] have been implicated in the pathogenesis of anastomotic leaks. The anastomotic leak rate in our current series of 19% is comparable with that reported in other series [3, 8 15]. We also noted a statistically significant increase in leak rates among those infants with wide-gap anomalies who underwent staged repairs as compared with those infants who underwent primary esophageal repairs (54% versus 11%; p 0.004). Although gap length between the pouches and anastomotic tension are important factors associated with anastomotic morbidity, the more likely explanation for the increased leak rate in our series is that 8 of 11 patients (72%) who underwent staged definitive repairs had colon interpositions. We noted a significantly higher leak rate among patients who underwent EC anastomoses versus EE anastomoses (75% versus 11%; p ). Anastomotic leak rates associated with colon interposition range from 9% [18] in one series to as high as 70% [19]. Another large series encompassing 227 patients reports a complication rate of 68% with colon interposition [3]. The high incidence of leak associated with cervical EC anastomoses has been attributed to impaired blood supply in the cephalad end of the interposition graft [18]. Additionally, leaks associated with colon interposition had a high reoperation rate. Five of 6 esophagocolonic leaks (83%) required surgical revision for intractable strictures or graft necrosis. Our stricture rate of 40% is slightly higher than that reported in other series [3, 8 14, 16]. We attribute this to the fact that a stricture was defined in our series as an anastomotic narrowing seen on UGI series that required one or more dilations. Other series define strictures as those requiring greater than two [7], three [14], or four [3, 8, 10] dilations with stricture rates ranging from 15% to 35% in these series [3, 7, 8, 10, 14]. Nine patients (15%) in our series required more than two dilations. A number of predisposing factors have been implicated in the pathogenesis of anastomotic strictures including a two-layer anastomosis [7, 10], gap length [8, 14], anastomotic tension [8, 14], silk sutures [13, 16], anastomotic leak [13, 15, 16], and GER [3, 7, 9, 16, 19]. In 9 of 12 patients (75%) with anastomotic leaks in our series, subsequent strictures developed that required dilation. Chittmittrapap and associates, in their series of 199 patients [15], noted an increased stricture rate in those patients with anastomotic leaks (71%) versus those without leaks (36%). The stricture rates in our series for EC anastomoses versus EE anastomoses were 50% versus 39%, a difference that was not statistically significant. However, all EE strictures were managed successfully with dilations, whereas 3/4 EC strictures required anastomotic revision. We currently favor the use of Vicryl (Ethicon, Somerville, NJ) sutures as well as extensive mobilization of the upper esophageal pouch routinely, to minimize anastomotic tension. If excessive tension persists despite upper pouch mobilization we advocate performing circular myotomies as described by Livaditis [20], mobilizing the gastroesophageal junction, or performing a Collis gastroplasty [21] in selective cases. Spitz [22] has advocated prophylactically paralyzing and ventilating infants who had anastomoses performed with extreme tension for 5 days postoperatively to minimize disruptive anastomotic forces. Although there were no leaks in that series, the stricture rate was 72% and the rate of GER was 54%. Chittmittrapap and associates [16] showed that GER significantly increased the stricture rate. Another large series of 303 patients noted a 52% rate of GER in 87 patients with anastomotic stricture [9]. In our series 5 of 25 patients (20%) with strictures had associated GER. Our GER rate (11%) is lower than rates quoted in other series of 37.9% [7], 39% [19], and 58% [3]. We believe that the true incidence of GER is higher than that reflected in our series, and that the disparity arises from the difficulty associated with properly documenting and confirming the presence of GER during postoperative follow-up

5 782 TSAI ET AL Ann Thorac Surg ESOPHAGEAL ATRESIA AND TRACHEOESOPHAGEAL FISTULA 1997;64: visits. We advocate aggressive diagnosis and treatment of GER to reduce postoperative stricture rates The reported incidence of recurrent TEF ranges between 3% [3] and 12% [13]. The incidence of recurrent TEF in our series was 10%. The pathogenesis of recurrent TEF has been an anastomotic suture line leak with erosion through the site of previous repair of the TEF [3]. Because recurrent TEFs rarely close spontaneously, we advocate reoperation with interposition of a pleural or pericardial flap after fistula division. Missed fistulas occurred in 2 patients (2.5%) in our series. One of these patients died of recurrent bronchopulmonary pneumonia and sepsis 6 months postoperatively. When the upper esophageal pouch cannot be distended with air on plain roentgenograms, nonionic water-soluble contrast medium should be used to opacify the proximal pouch, so that proximal fistulas may be recognized preoperatively. It is also extremely important to routinely dissect the proximal pouch to the thoracic inlet and to completely divide the common wall between the trachea and proximal esophagus to avoid overlooking proximal fistulas. One needs to have a high level of suspicion for this complication postoperatively because the symptoms associated with missed fistulas, such as aspiration and recurrent respiratory tract infections, may be attributed to GER or tracheomalacia. When recurrent or missed fistulas are strongly suspected clinically, cineesophagography may need to be repeated several times. Special techniques such as placing the patient in a prone position during contrast esophagography and injecting contrast medium at 2-cm increments while gradually withdrawing the nasogastric tube may help diagnose these entitities. If contrast studies have negative results we advocate simultaneous esophagoscopy and bronchoscopy with methylene blue injection or cannulation of the fistula. The association of tracheomalacia with EA and TEF has been well recognized, with a reported incidence ranging from 16% [13] to 33% [23]. The incidence of tracheomalacia in our series is 11%. The most serious problem associated with tracheomalacia is dying spells, which occurred in 2 patients in our series, who required aortopexy and stent placement respectively, for control of symptoms. When severe tracheomalacia is documented by bronchoscopy in symptomatic infants (apnea, cyanosis, recurrent pneumonia with expiratory obstruction of the trachea) surgical intervention should not be delayed. The management of newborns with isolated EA (type A) and long-gap type C anomalies is often complex and presents a challenge for pediatric surgeons. The operative morbidity associated with esophageal replacement procedures is high [3, 18, 19, 24]. In addition to anastomotic leaks, strictures, and graft ischemia, other late complications include GER, recurrent aspiration, esophageal dysmotility, swallowing problems, ulceration, bleeding, respiratory tract infections, tortuosity of the graft, and failure to thrive [3, 18, 24]. The increased morbidity has served as an impetus for developing new methods for bridging the gap between the esophageal segments for wide gap anomalies. Howard and Myers [25] first reported upper pouch bougienage with mercury-filled bougies as a method of elongating the upper pouch. Others have advocated esophageal autoanastomosis by producing a mucosalined fistula [26], lengthening of the upper pouch by circular myotomy [20], and electromagnetic esophageal bougienage [27]. Puri and colleagues [28] demonstrated that mechanical stretching of the pouches is not necessary because spontaneous growth and hypertrophy of the esophageal pouches occurs in the first months of life. Bensoussan and associates [29] reported 6 infants with wide-gap anomalies who underwent primary repair at 3 months of age after spontaneous elongation of the esophagus. Kimura and Soper [30] reported multistaged extrathoracic esophageal elongation in patients with cervical esophagostomies in an attempt to elongate the native esophagus and perform a primary repair. A review of the data obtained from several large series [8, 14, 24, 28, 31, 32] of patients with long-gap anomalies who underwent delayed EE anastomoses revealed leak rates that varied widely from 0% [32] to 100% [14] and stricture rates that ranged from 50% [8, 32] to 90% [24]. The high morbidity in these cases has been attributed to anastomotic tension. Despite the high complication rate reported for delayed EE anastomoses for long-gap anomalies, most authors [8, 14, 24, 28, 31, 32] favor primary EE anastomoses as compared with esophageal replacement. The long-term functional results were much better with EE anastomoses as compared with esophageal replacement procedures [24, 31, 32] (ie, dysphagia and feeding difficulty were associated with esophageal substitution). Additionally, despite the high incidence of anastomotic strictures associated with EE anastomoses in these patients [8, 14, 24, 28, 31, 32], most of these strictures responded to dilation and very few required surgical revision, as was the case in our series. In Rescorla and associates series [24], 2/4 EC strictures (50%) required surgical revision as compared with 2/9 EE strictures (22%). The overall incidence of secondary surgical procedures for colon grafts was also higher [24] when compared with the reoperation rate for EE anastomoses. The reoperation rate among colon grafts was also higher in our current series. In addition to leaks and strictures, operative indications included revision for colon redundancy [24], bleeding from ulceration [24], and colon fibrosis [14]. Life-threatening complications (eg, colon graft necrosis) occurred with a higher frequency among those patients who underwent EC versus EE anastomoses [24], as was the case in the present series. Because of our high complication rate with colon interposition in patients with wide-gap anomalies, we advocate expectant management with suctioning of the upper pouch and feeding through gastrostomy tubes while awaiting natural elongation of the esophagus. Delayed primary anastomosis can be performed between 8 and 12 weeks as advocated by Puri and associates [28] and Bensoussan and colleagues [29]. In cases of preexisting cervical esophagostomy, the technique advocated by Kimura and Soper [30] of extrathoracic esophageal elon-

6 Ann Thorac Surg TSAI ET AL 1997;64: ESOPHAGEAL ATRESIA AND TRACHEOESOPHAGEAL FISTULA 783 gation seems promising and may offer an alternative to esophageal replacement. Although the morbidity associated with surgical repair of those anomalies is high, this does not affect the overall survival. The high complication rate associated with colon interposition suggests that one should preserve the native esophagus as a primary conduit whenever feasible. If one excludes infants with major associated lifethreatening congenital anomalies, survivals exceeding 90% can be achieved. References 1. Haight C, Towsley HA. Congenital atresia of the esophagus with tracheoesophageal fistula: extrapleural ligation of fistula and end-to-end anastomosis of esophageal segments. Surg Gynecol Obstet 1943;76: Poenaru D, Laberge JM, Neilson IR, Guttman FM. A new prognostic classification for esophageal atresia. Surgery 1993;113: Engum SA, Grosfeld JL, West KW, Rescorla FJ, Scherer TR. Analysis of morbidity and mortality in 227 cases of esophageal atresia and/or tracheoesophageal fistula over two decades. Arch Surg 1995;130: Redo SF. Congenital esophageal atresia and tracheoesophageal fistula fifteen year experience. NY State J Med 1975;75: Gross RE. The surgery of infancy and childhood. Philadelphia: Saunders, Waterston DJ, Bonham-Carter RE, Aberdeen E. Oesophageal atresia: tracheo-oesophageal fistula: a study of survival in 218 infants. Lancet 1962;1: Manning PB, Morgan RA, Coran AG, et al. Fifty years experience with esophageal atresia and tracheoesophageal fistula. Ann Surg 1986;204: McKinnon LJ, Kosloske AM. Prediction and prevention of anastomotic complications of esophageal atresia and tracheo-esophageal fistula. J Pediatr Surg 1990;25: Spitz L, Kiely E, Brereton RJ, Drake D. Management of esophageal atresia. World J Surg 1993;17: Holder TM, Cloud DT, Lewis JE, et al. Esophageal atresia and tracheoesophageal fistula: a survey of its members by the surgical section of the American Academy of Pediatrics. Pediatrics 1964;34: Bishop PJ, Klein MD, Phillipart AI, et al. Transpleural repair of esophageal atresia without a primary gastrostomy: 240 patients treated between 1951 and J Pediatr Surg 1985; 20: Touloukian RJ. Long-term results following repair of esophageal atresia by end-to-side anastomosis and ligation of the tracheoesophageal fistula. J Pediatr Surg 1981;16: Spitz L, Kiely E, Brereton RJ. Esophageal atresia: five year experience with 148 cases. J Pediatr Surg 1987;22: Sillen U, Hagberg S, Rubenson A, Werkmaster K. Management of esophageal atresia: review of 16 years experience. J Pediatr Surg 1988;23: Chittmittrapap S, Spitz L, Kiely EM, Brereton RJ. Anastomotic leakage following surgery for esophageal atresia. J Pediatr Surg 1992;27: Chittmittrapap S, Spitz L, Kiely EM, Brereton RJ. Anastomotic stricture following repair of esophageal atresia. J Pediatr Surg 1990;25: Louhimo I, Lindahl H. Esophageal atresia: primary results of 500 consecutively treated patients. J Pediatr Surg 1983;18: DeLorimer A, Harrison MR. Esophageal replacement. In: Aschraft KW, Holder TM, eds. Pediatric esophageal surgery. Orlando: Grune & Stratton, 1986: Holder TM, Ashcraft RW, Sharp RJ, Amoury RA. Care of infants with esophageal atresia, tracheoesophageal fistula, and associated anomalies. J Thorac Cardiovasc Surg 1987;94: Livaditis A. Esophageal atresia. A method of overbridging large segmental gaps. Z Kinderchir 1973;13: Evans M. Application of Collis gastroplasty to the management of esophageal atresia. J Pediatr Surg 1995;30: Spitz L. Esophageal atresia: past, present, and future. J Pediatr Surg 1996;31: Slany E, Holzki J, Holschneider AM, et al. Flaccid trachea in tracheo-oesophageal malformations. Z Kinderchir 1990;45: Rescorla FJ, West KW, Schere LR, Grosfeld JL. The complex nature of type A (long-gap) esophageal atresia. Surgery 1994; 116: Howard R, Myers NA. Esophageal atresia: a technique for elongating the upper pouch. Surgery 1965;58: Shafer AD, David TE. Suture fistula as a means of connecting upper and lower segments in esophageal atresia. J Pediatr Surg 1974;9: Hendren WH, Hale JR. Esophageal atresia treated by electromagnetic bougienage and subsequent repair. J Pediatr Surg 1976;11: Puri P, Ninan GK, Blake NS, Fitzgerald RJ, Guiney EJ, O Donnell B. Delayed primary anastomosis for esophageal atresia: 18 months to 11 years follow up. J Pediatr Surg 1992; 27: Bensoussan AL, Letourneau JN, Blanchard H. Cure de l atresie de l oesophage (de types 1 et 2). Anastomose primaire differee. Chir Pediatr 1983;24: Kimura K, Soper RT. Multistaged extrathoracic esophageal elongation for long gap esophageal atresia. J Pediatr Surg 1994;29: Ein SH, Shandling B, Heiss K. Pure esophageal atresia: outlook in the 1990s. J Pediatr Surg 1993;28: Boyle EM, Irwin ED, Foker JD. Primary repair of ultra-longgap esophageal atresia: results without a lengthening procedure. Ann Thorac Surg 1994;57: INVITED COMMENTARY Doctor Tsai and her coauthors have presented a 20-year experience with a relatively small series of infants and children with esophageal atresia and tracheoesophageal fistula (EA/TEF). They have reported on 81 patients of whom 68 underwent definitive treatment of their anomaly. Their overall survival rates are excellent, with favorable cases (Waterston risk groups A and B) achieving a survival between 89% and 100%. Although the Waterston classification has been traditionally used to categorize infants with EA/TEF, the criteria are really out of date in the 1990s; the new classification reported by Spitz and associates [1] in 1991 is more appropriate today. The only criteria of importance in this new classification are weight of more or less than 1,500 g and the presence or absence of severe cardiac anomalies. Although the survival rates in this series are comparable with those seen in other, current, large series, such as those reported by Spitz, Myers, and myself, the complication rates in this series are significantly higher. For instance, in a series of almost 600 patients at the Univer by The Society of Thoracic Surgeons /97/$17.00 Published by Elsevier Science Inc PII S (97)