Tracheal Compression as a Cause of Apnea Following Repair of Tracheoesophageal Fistula: Treatment by Aortopexy

By Marshall Z. Schwartz and Robert M. Filler

Galveston, Texas and Toronto, Ontario, Canada

O Tracheal compression is one :of the many causes 0f apnea following repair of esophageal atre- sia with tracheoesophageal fistula. Since May, 1977 w e have treated eight patients with "dying spells'" usually occurring during or following a feeding. These episodes were characterized by cyanosis and bradycardia, which progressed to apnaa. Tracheal compression was documented by bronchoscopy in all cases and determined to be the etiology of the apnea spells. Other possible causes such as aspira- tion from gastroesophageal reflux, recurrent tra- cheoesophageal f is tu la and esophageal stricture were eliminated. Eight infants underwent aortopexy that was successful in terminating the apneic episodes in seven patients with an average follow- up of 20 roD. One patient continued to have symp- toms from a previously undiagnosed vascular ring which was successfully divided one month following aortopexy. After being asymptomatic for 1 mo this infant died from a cardiorespiratory arrest at home. An autopsy did not reveal the cause. Intermittant tracheal obstruction leading to apneic spells is a life-threatening entity following tracheoesophageal fistula repair, which can be corrected by aortopexy.

INDEX WORDS: Esophageal atresia; tracheoesopha- geal fistula; apnaa; aortopexy; tracheal compres- sion.

T R A C H E A L compression is one of the many causes of respiratory symptoms following

repair of esophageal atresia and tracheoesopha- geal fistula. In these cases the trachea, which may be more pliable than normal, is compressed between th6innominate artery or aorta and the esophagus. Tracheal compression can produce only mild symptoms and escape diagnosis, or it

From the Child Health Center, University o f Texas Medi- cal Branch, Galveston, Texas, The Hospital for Sick Chil- dren, Torontol Ontario, Canada, and Children's Hospital Medical Center, Boston, Mass.

Presented before the 1 l th Annual Meeting o f the Ameri- can Pediatric Surgical Association, Marco Island, Florida, May 7-10, 1980.

Address reprint requests to Marshall Z. Schwartz, M.D., Department o f Surgery, University o f Texas Medical Branch, Galveston, Texas 77550.

�9 1980 by Grune & Stratton, Inc. 0022-3468/80/1506~)02l $01.00/0

may result in a dramatic symptom complex often referred to as "dying spells." T h e s e episodes, which usually do not appear until 2 or 3 m0 o f age, are characterized by cyanosis rapidly progressing to apnea and most often occur during or shortly after a feeding. The recognition of tracheal compression as the cause o f apneic spells can be perplexing because similar symp- toms can be caused by aspiration due to esopha- geal stricture, recurrent tracheoesophageal fiStU- la, or gastroesophageal reflux; ,all well recog- nized sequelae of esophageal atresia repair. In a previous report the symptom complex, methods of diagnosis, and treatment by an aortopexy were described in three infants.! This study adds an additional eight cases and attempts to further clarify this entity.

MATERIALS AND METHODS

From May 1977 to August 1979 eight infants who had had repair of esophageal atresia were admitted to the Hospital for Sick Children in Toronto or the Children's Hospital Medical Center in Boston for treatment of apneic spells due to tracheal compression.

Each of the 3 females and 5 males in this group were born with esophageal atresia and a tracheoesophageal fistula to the distal esophagus. The birth weight was below 2100 g in 6 infants and below 1800 g in 3 infants. Associated anomalies are shown in Table 1. Initial surgery was performed at 1 of the above institutions in 5 cases and elsewhere in 3 cases. Repair by division of the fistula and esophagoesophagostomy was performed in the first 2 days of life in 5 patients. Complete repair was delayed from 6 to 11 weeks in three premature infants treated initially by gastrostomy and suction to the upper pouch. In 1 of these patients (case 1), the fistula was divided at 2 wk of age and gastrostomy feedings started: In the other 2 patients (cases 2 and 4), nutrition was provided parenterally prior to esophagoesophagostomy. Seri- ous postoperative complications were noted in two cases. Case 5 developed a recurrent tracheoesophageal fistula 1 week following repair that was successfully closed 5 wk later. This infant also developed hypertrophic pyloric stenosis at 8 wk of age requiring pyloromyotomy. Case 4 developed respi- ratory obstruction on the third day following repair of esophageal atresia due to a right vocal cord paralysis. He required a tracheostomy for 8 wk. Except for case 5, none of. the patients had an anastomotic leak or esophageal stricture.

Apneic episodes began 6 wk-6 mo (average = 3.5 mo) after repair of esophageal atresia and traeheoesophageal

842 Journal of Pediatric Surgery, Vol. 15, No. 6 (December), 1980

ESOPHAGEAL ATRESIA

Table 1. Clinical Data Relating to Treatment of Esophageal Atresia and Tracheoesophageal Fistula

843

Age at Birth Definitive

Weight Significant Associated Operation Patient (g) Anomalies (wk) Complications

1 1,100 Atrial septal defect 11

2 1 ,450 6

3 1,770 NB

4 2 ,000 Imperforate anus 10

Vesicoureteral reflux

5 2 ,040 NB

6 2 ,065 Right aortic arch NB

Left lig. arteriosum

Anomolous left

subclavian artery

7 2 ,270 Imperforate anus NB

8 3 ,500 NB

Right vocal cord

paralysis*

Recurrent TEF t

Pyloric stenosis~

NB = Newborn

*Required tracheostomy for 2 too.

tRepair at age 6 wk.

:l:Repair at age 2 wk.

fistula and when the infants were out of the hospital (Table 2). The spells were associated with cyanosis and bradycardia and they usually occurred during or immediately after feed- ing. Resuscitation at home or in the hospital was necessary in each case. As part of the initial investigation of these spells, each infant had a barium esophagram, and a neurologic and cardiac evaluation. In two patients, lower esophageal sphincter manometry and a technetium reflux scan were performed to more accurately assess the likelihood of gastro- esophageal reflux as the cause of symptoms (Table 3). After these studies suggested that tracheal compression was the likely cause of apneic spells, confirmation was obtained by bronchoscopy under general anesthesia.

All infants were treated by aortopexy at the time or shortly after the initial bronchoscopy. During surgery inhalation anesthesia was administered through a standard endotra- cheal tube or, more recently through a ventilating broncho- scope. Thus, relief of tracheal compression could be assessed as the corrective operation was proceeding.

The aim of surgery was to relieve pressure on the trachea by suspending the anterior vascular structures from the

under surface of the sternum. The operative procedure was adapted from the technique originally described by Gross 2 and subsequently described by others. 1 A left anterior thora- cotomy through the third interspace was used to gain access to the anterior mediast inum except in case six who was approached from the right chest because of a right aortic arch. In two cases the dissection was done extrapleurally to avoid violation of the pleural cavity and, therefore, avoid the need for a transpleural chest tube. The left (or right) lobe of the thymus was excised to expose the aortic arch and its branches and to create a space for the translocation of these vessels. The tissue plane between the aorta and trachea was not dissected since it is this at tachment that pulls the anterior wall of the trachea forward when the vessels are suspended from the sternum.

Three to four 3-0 nonabsorbable sutures were placed through the adventitia and a portion of the media of the vessels at the following locations: (1) aortic arch, at the level of the lateral border of the innominate artery; (2) anterior wall of the innominate artery, 0.5-1 cm from its origin; (3) aortic arch, at the level of the medial border of the innomi-

Table 2. Clinical Data Relating to the Onset and Treatment of Apnea Spells

Age at Onset Age at Aortopexy Follow-Up (mo) (mo) Complications (too) Result

1 4 4 .5 - - 28 Excellent

2 4 9 - - 12 Excellent

3 2 11 - - 24 Excellent

4 5 29 - - 24 Excellent

5 5 6 - - 7 Excellent

6 1.5 3.5 Recurrent symptoms 2 Dead

due to vascular

r ing*

7 6 8 - - 28 Excellent

8 3 5 - - 15 Excellent

*Repaired at 4 too,

844 SCHWARTZ AND FILLER

Table 3. Results of Studies Used to Evaluate Patients With Apnea Spells

Case Stricture

Barium Swallow

Aspiration

Lower Esophageal Tracheal Technitium Reflux Sphincter

Narrowing Other Scan Manometry Bronchoscopu

+ Gastroesophageal TC

reflux + Gastroesophageal TC

reflux

+ TC

+ TC

+ TC,

+ TC

+ Normal Normal TC

+ Normal Normal TC

TC = Tracheal compression,

nate artery; and (4) aortic arch, 0.5 cm medial to the previous suture. A plane was then created along the anterior surface of the s ternum and both ends of the previously placed sutures were passed through the sternum and tied in sequence. This maneuver elevated that portion of the innomi- nate artery and aorta that impinged on the trachea and pulled the trachea away from the esophagus. When a bronchoscope was being used to maintain anesthesia, relief of the anterior compression on the trachea could be assessed immediately. Otherwise, bronchoscopy was performed at the completion of the procedure to evaluate the effect of aor- topexy. The chest wound was then closed and the pleural cavity drained by underwater suction, The drainage catheter was removed in 24 hr.

RESULTS

The interval from the onset of spells to defini- tive diagnosis and treatment varied from 2 wk to 2 yr. In 3 infants, the interval was greater than 2 too. In 2 of these 3 patients (cases 2 and 3) a delay of 5 mo and 9 mo, respectively, was due to failure to associate the symptom complex with tracheal compression. In the other child (case 4) the spells were so widely spaced that surgery was postponed until the attacks became more frequent and it became obvious that the tracheal compression would not resolve spontaneously.

Diagnostic studies ruled out the presence of recurrent tracheoesophageal fistula and signifi- cant gastroesophageal reflux in each case. No child had an esophageal stricture and neither neurologic nor cardiac causes for the spells could be found. In all the patients radiographic evidence of tracheal compression was best seen during the expiratory phase of respiration by examining the lateral view of the air filled trachea at fluoroscopy or on plain radiographs of the neck and chest (Figs. 1 and 2). Bronchoscopy revealed significant obliteration of the tracheal

lumen by an anterior pulsating mass in all cases. This was associated with bulging of the posterior tracheal wall into the lumen in most patients (Fig. 3). The bronchoscope could be advanced beyond the collapsed segment without difficulty. After aortopexy, bronchoscopy revealed a marked increase in the antero-posterior diameter of the tracheal lumen.

There were no immediate complications following aortopexy. In seven patients apneic spells ceased and these children have remained without spells from 7 to 28 mo (average=20 mo) following aortopexy. In case 6, apneic spells recurred 1 wk following suspension of a right aortic arch. Preoperative barium swallow did not show a posterior indentation on the esophagus indicative of a vascular ring, nor had a vascular

Fig. 1. Compression of air filled trachea observed on lateral film of the neck and chest at age 5 me (case 5).

ESOPHAGEAL ATRESIA 845

Fig. 2. Caliber of air filled trachea as it is affected by swallowing (case 2). (A) During a barium meal the trachea is not collapsed when esophagus is relatively empty; (B) Distention of the esophagus causes tracheal compression,

ring been noted at the time of esophageal atresia repair or aortopexy. However, in evaluating the recurrent spells after aortopexy, a barium swal- low did show the presence of a structure behind the esophagus and angiography confirmed the presence of an anomalous left subclavian artery and left ligamentum arteriosum completing a vascular ring. Thus, suspension of the right aortic arch had relieved the anterior tracheal compression, but when the entire ring was displaced anteriorly, its posterior portion pro- duced tracheal compression. After division of the anomalous artery and ligamentum arteriosum, apneic spells disappeared. While at home 1 mo later, and 30 min after a meal, this child suddenly became apneic and cyanotic and could

not be resuscitated. An autopsy was performed, but no cause of death could be found.

DISCUSSION

The association between anterior tracheal compression and ventilatory difficulty was first described by Gross and Neuhauser in 1948. 2 Subsequently, Filler et al., 1 Fearon et al., 3 Mustard et al., 4 and Berdon et al. ~ noted that a significant number of their patients who presented with tracheal obstruction due to compression by the innominate artery were infants who had had prior repair of esophageal atresia and tracheoesophageal fistula. These authors indicated that symptoms of airway obstruction could be provoked by any event that

846 SCHWARTZ AND FILLER

Fig. 3. Appearance of distal trachea at bronchoscopy in child with tracheal collapse. (A) Before aortopexy--slit like opening; [B) A l ta r aortopexy--relatively normal caliber airway.

increased upper mediastinal crowding, such as drinking.

Other investigators have also recognized the association between infants born with tracheo- esophageal anomalies and intermittant tracheal obstruction. 6'7 These authors suggest that an abnormally pliable trachea rather than mediasti- nal crowding might be the primary underlying factor which leads to tracheal collapse, airway obstruction and apneic spells. Benjamin et al. 6 found that 21 of 80 children with esophageal atresia developed respiratory symptoms follow- ing repair. At bronchoscopy 20 of these symp- tomatic patients had "tracheomalacia" that was defined as a generalized or localized weakness of the tracheal wall. 6 Davies and Cywes 7 described one patient with a "flaccid trachea" who required prolonged hospitalization because of several episodes of respiratory arrest associated with feedings but not felt to be due to aspiration. Meradji et al. 8 screened 63 children radiographi- cally with corrected esophageal atresia and noted a "soft trachea" in 31 patients (50%).

Until recently there has been scant histologic data to support the hypothesis of a structural defect in the trachea. Emery and Haddadin 9 reported finding extensive squamous metaplasia of the tracheal mucosa in 40 of 50 children dying with esophageal atresia and tracheoesophageal fistula. Wailoo and Emery ~~ extended this work in reporting on a remarkable unselected sequen- tial series of tracheas obtained from 53 deceased patients with esophageal atresia and tracheo- esophageal fistula. The ages of these children ranged from 28 wk gestation to 2.5 yr. In 40 of these specimens, transverse sections were exam-

ined histologically at 6# intervals. The total internal perimeter of the trachea and the relative proportions of this perimeter consisting of carti- lage and muscle were determined. Their evalua- tion showed that in 75% of the tracheas there was a segment in which the length of circular cartilage was decreased with a concomitant increase in the length of muscle in the membra- nous portion of the tracheal ring. Wailoo and Emery concluded that the effect of a decrease in the cartilage to muscle ratio resulted in a greater tendency for antero-posterior collapse. 9 Since these tracheas were taken from patients who succumbed from respiratory problems, the sever- ity and number of abnormalities might be less in those infants who ultimately survive. Neverthe- less, these findings provide objective evidence to indicate that the trachea in children with esophageal atresia is often abnormally pliable making it readily susceptible to collapse by external compression.

Possible reasons for the abnormal tracheal development in these infants were discussed by DavieS a n d Cywes. 7 Since the trachea and esophagus are derived from the primitive fore- gut, the trachea lies closely applied to - the esophagus during early development. As noted, these structures as well as the aortic arch and great vessels are confined to a relatively small space in the superior mediastinum. These authors surmise that esophageal dilatation above the site of atresia compresses the fetal trachea and prevents its normal development. A second hypothesis proposes that abnormal tracheal development is due to a loss of intratracheal pressure in the fetus] The embryonic tracheo-

ESOPHAGEAL ATRESIA 847

bronchial tree is normally filled with fluid and in the presence of a competent glottic mechanism this fluid provides internal support for the trachea. However, if the trachea can be decom- pressed through a tracheoesophageal fistula the splinting effect may belost allowing the trachea to be more susceptible to the effects of external pressure:

The etiology of apneic spells in children following repair of esophageal atresia and tracheoesophageal fistula can be elusive because neurologic disorders, tracheal obstruction, or aspiration due to recurrent tracheoesophageal fistula, poor esophageal emptying, or gastro- esophageal reflux, can be responsible. Barium swallow is useful to evaluate the likelihood of aspiration due to tracheoesophageal fistula or gastroesophageal reflux. Since most of these children have motor disorders of the lower esophagus, it is often difficult to decide if these abnormalities are responsible for the spells. This is especially true when these findings are not marked or when a degree of tracheal compres- sion is also identified. The technitium reflux scan and esophageal manometry may give additional insight into the mechanics of esophageal motility but may not be able to specifically determine the etiology of the apnea episodes. Twenty-four-hour esophageal pH monitoring may be useful if a fall in pH can be correlated with respiratory symp- toms. 1 t

Assessment of the caliber of the air filled trachea at the time of barium swallow is useful to evaluate the possibility of tracheal compres- sion as the cause of apnea. A mild degree of airway narrowing is common in all children who have had repair of esophageal atresia but severe narrowing is usually significant. I In our experi- ence bronchoscopy represents the definitive diagnostic tool in evaluating tracheal collapse. In all of our cases marked tracheal compression was obvious on direct vision (Fig. 3). Bronchoscopy is indicated in all children who develop apneic spells after esophageal atresia repair whenever radiographic studies suggest that tracheal narrowing is present.

It is well recognized that tracheal compression may be associated with stridor, cyanosis, and recurrent pneumonia but it is not commonly appreciated that compression resulting in apnea may occur intermittently in a child who is not

symptomatic between episodes. In fact, apnea spells were the only significant respiratory mani- festation of tracheal compression in 10 of our total experience of I 1 patients (3 from a previous report~).

It has been proposed that a reflex mechanism although not the initiating factor, is responsible for the apneic spells. This reflex may be mediated by the vagus nerve and triggered when the trachea is stimulated whether internally by tracheal secretions which cannot be expecto- rated, or externally by a bolus of food in the esophagus.

Of eight children, 6 in this series and 1 of the 3 in our previous series were born prematurely with birth weights ranging between 1450 and 2100 g. In 4 of these 11 cases repair of esopha- geal atresia was delayed. The incidence of prematurity, low birth weight, and the need for delayed surgery is higher in this group than would be expected in an unselected group of infants with esophageal atresia. 12 However; the relationship between prematurity and the devel- opment of tracheal compression is unknown. Another as yet unexplained finding from our data is that apneic spells did not begin for one to six months following esophageal atresia repair.

Tracheal compression can be treated by several techniques. Surgical intervention proba- bly is unnecessary when the respiratory symp- toms are not severe and there is no history of apneic spells. Usually, by one year of age mild respiratory symptoms will resolve because of an increase in the rigidity of the trachea. However, one case in this series eventually required surgery at 21/2 yr of age because of infrequent but persistent spells. Some authors have suggested prolonged endotracheal intubation for moderate to severe Symptoms. 7'8 Szarnicki et al. 13 reported a patient who underwent placement of a dacron graft between the ascending and descending aorta with division of the aortic arch in a 10- mo-old child to relieve tracheal compression which developed after right pneumonectomy. In our opinion these techniques subject infants with tracheal compression to significant morbidity and mortality and in general should be rejected in favor of aortopexy. The aim of this uncompli- cated procedure is to provide more space in the crowded upper mediastinum and reduce external pressure on what is probably a more compress-

848 SCHWARTZ AND FILLER

able tracheal wall. This technique obviously does not alter any s tructural deficit in the wall of the trachea.

The safety and efficacy of aortopexy for the t rea tment of tracheal compression appears to be

well established. ~,4,5 We believe that this proce- dure is indicated for those infants who have sustained one or more l i fe-threatening episodes of apnea, and in whom the diagnosis of tracheal compression is confirmed by bronchoscopy.

R E F E R E N C E S

1. Filler RM, Rosselo P J, Lebowitz RL: Life-threatening anoxic spells caused by tracheal compression after repair of esophageal atresia: Correction by surgery. J Pediatr Surg 11:739-748, 1976

2. Gross RE, Neuhauser EB: Compression of the trachea by an anomalous innominate artery. An operation for its relief. Am J Dis Child 75:57(~574, 1948

3. Fearon B, Shortreed R: Tracheobronchial compression by congenital cardiovascular anomalies in children. Syndrome of apnea. Ann Otolaryngol 72:949 969, 1963

4. Mustard WT, Bayliss CE, Fearon B, et al: Tracheal compression by the innominate artery in children. Ann Thorac Surg 8:312-319, 1969

5. Berdon WE, Baker DH, Bordiuk J, et al: Innominate artery compression of the trachea in infants with stridor and apnea. Radiology 92:272-278, 1969

6. Benjamin B, Cohen D, Glasson M: Tracheomalacia in association with congenital tracheoesophageal fistula. Sur- gery 79:504 508, 1976

7. Davies MR, Cywes S: The flaccid trachea and

tracheoesophageal congenital anomalies. J Pediatr Surg 13:363-367, 1978

8. Meradji LW, Van Walleghem J, Vervat D: A radiolog- ical study of 63 cases of operated esophageal atresia. Radiol Clin Biol 43:559-569, 1974

9. Emery JL, Haddadin A J: Squamous epithelium in respiratory tract of children with tracheoesophageal fistula. Arch Dis Child 46:236-242, 1971

10. Wailoo MP, Emery JL: The trachea in children with tracheoesophageal fistula. Histopathology 3:329 338, 1979

11. Jolley SG, Johnson DG, Herbst J J, et al: An assess- ment of gastroesophageal reflux in children by extended pH monitoring of the distal esophagus. Surgery 84:16-24, 1978

12. Waterston D J, Bonham-Carter RE, Aberdeen E: Oesophageal atresia: Tracheoesophageal fistula. A study of survival in 218 infants. Lancet 1:819, 1962

13. Szarnicki R, Maurseth K, DeLeval M, et al: Tracheal compression by the aortic arch following right pneumonec- tomy in infancy. Ann Thorac Surg 25:231-235, 1978

D i s c u s s i o n

L. Leape (Boston, Massachusetts): I want to thank Dr. Filler for br inging us up-to-date on a

very impor tant complication after repair of esophageal atresia. I want to recall one pat ient who illustrates some of the problems we get into. This was a child who had two l ife-threatening apneic spells requiring full resuscitation, one of them in the hospital, which were found to be due to gastroesophageal reflux. This was corrected. He had no further spells, but then went on to develop episodes of severe respiratory distress which turned out to be due to compression of his trachea from the esophagus. These were relieved by aortoplexy. I merely ment ion it to indicate that a pat ient can have more than one cause for apneic spells.

K. Koenigsberg (New York, New York): I 'd like to ask Dr. Filler whether on that last child

who died if there had been any studies of the EKG while the child had been swallowing.

R. M. Filler (closing): I 'd like to thank Dr. Leape for his remarks. I might just add one extra note about these spells. I ment ioned that there is the theory that suggests that these spells are due to a vagally triggered reflex. But as one watches these children, one really does get the impression that progressive anoxia develops while eating. The infant appears to be so hungry that he keeps gu lp ing despi te the fac t tha t swal lowing increases airway obstruction. As a result hypoxia increases and bradycardia, cardiac arrest, and respiratory arrest ensue. In answer to Dr. Koenigsberg's question an E K G was done in all of these cases, but an E K G was not done dur ing eating or during swallowing.