48: Management of Esophageal Perforation

Esophageal perforation is a challenging clinical condition that requires prompt diagnosis and management. Delay in identification and/or treatment results in high rates of morbidity and mortality. There are sharp differences in etiology, presentation, treatment, and outcome of cervical versus thoracic perforation of the esophagus. Most cervical perforations respond well to simple drainage alone. Although the treatment of thoracic esophageal perforations is individualized, most patients are candidates for primary repair regardless of the time to presentation or intervention. Improvements in endoscopic stent technology and increased experience with placement support this modality as a viable treatment option in cases of benign, malignant, and iatrogenic esophageal perforation.

Esophageal perforation usually is the result of iatrogenic injury caused by instrumentation (e.g., esophagoscopy, bougienage, and achalasia dilation)^1-3 (Table 48-1). The most common site for perforation of the normal esophagus is at its most proximal location, immediately above the cricopharyngeus muscle and below the inferior pharyngeal constrictor (Killian's triangle). Injury at this location is most often caused by attempted forceful intubation of the esophagus for endoscopy (rigid or flexible) in a patient who is not sufficiently anesthetized. Other common sites of perforation include those in which the esophagus is normally narrowed (the distal esophagus), pathologically narrowed, or anatomically abnormal. Occasionally, intramural perforation can occur when the mucosa is sheared off the muscularis during endoscopy or bougienage. These conditions are not perforations in the truest sense, but present in a similar fashion and must be differentiated from frank perforation. Spontaneous perforation is a misnomer; it is more accurately termed barogenic perforation (Boerhaave syndrome). Blunt and penetrating trauma contributes only a small number of perforations.⁴ Foreign bodies, infections, and operative injuries are additional, although rare, causes worth noting.⁵

Pain is the most common complaint in patients presenting with esophageal perforation. In addition, cervical perforations can present with dysphagia, odynophagia, and dysphonia. Subcutaneous emphysema is often palpable in the neck. Pain from an intrathoracic perforation may be localized initially to the subxiphoid region, and hence may be misinterpreted as a myocardial infarction, aortic dissection, perforated duodenal ulcer, or pancreatitis. The pain also may be substernal, referred to the back, or poorly localized, and is usually severe. Dyspnea and anxiety are common associated findings. Tachycardia is also common, and fever develops early in the clinical course. Shock is rarely seen after cervical perforation because of local containment, whereas with free perforation into the pleural space, as often occurs with Boerhaave syndrome, rapid progression to shock may occur within 24 hours. Perforation by balloon dilation of achalasia similarly may result in a rapid and disastrous course if treatment is delayed. A small instrumental perforation, on the other hand, confined to the wall of the esophagus or situated in the mediastinum, may lead to a much slower development of symptoms. The key feature in the diagnosis of all esophageal perforations is a high index of suspicion. Prompt evaluation of patients with any of these findings or symptoms after endoscopy is most important because early diagnosis improves outcome.

A lateral plain neck film typically demonstrates air in the prevertebral space even early after cervical perforation (Fig. 48-1). Chest radiographs may show a pleural effusion, subcutaneous emphysema, pneumomediastinum, or pneumothorax. Radiographs soon after the event may be normal, although 75% of patients have abnormal films within 12 hours of presentation. Examination of the esophagus with contrast material is preferably done with water-soluble medium (Figs. 48-2 to 48-5). If the diagnosis is unclear or the definition is inadequate, dilute barium may be used sparingly. The overall false-negative rate in this setting is at least 10%.³ If perforation is still strongly suspected, a swallow study should be repeated several hours later, or another modality of investigation might prove more helpful. CT scans can aid in the diagnosis of esophageal perforation and identify late sequelae⁶; evidence of perforation may include extraluminal air, extravasation of contrast material, esophageal thickening, and communication of the air-filled esophagus with adjacent structures (Fig. 48-6). Oral or nasogastric tube contrast material can be administered just prior to CT scan, increasing its sensitivity. Planning the diagnostic evaluation with a radiologist is recommended to avoid the administration of contrast material that may delay the use of CT. Endoscopy is occasionally helpful, either to diagnose a perforation (usually in the context of trauma) or to assess the status of the esophagus in planning operative repair. Important information that can be determined by endoscopy includes the exact location of the tear, the presence of a stricture, or the extent of a carcinoma. If carefully performed with a flexible endoscope by an experienced endoscopist/surgeon, there is no additional risk to performing esophagoscopy when potentially important information can be gained.

Intramural perforations can be a source of confusion to the inexperienced radiologist or surgeon. These are usually caused by instrumentation; the mucosa is inadvertently sheared off the muscularis at the cricopharyngeus or a pathologically narrowed area. Intense pain is a common symptom. Barium swallow frequently demonstrates two lumens: the true lumen in which contrast material readily passes downward and a false lumen with contrast pooled in a dependent pouch (Fig. 48-7). CT imaging can help to clarify this diagnosis (Fig. 48-8). Critical to the early detection of esophageal perforation is the physician's awareness of the possibility. Where instrumentation has preceded the development of any signs or symptoms, perforation must be suspected until proven otherwise.

Factors important to selecting the appropriate treatment include the location, etiology, and status of the native esophagus. In addition, the severity of contamination, often associated with delay in seeking treatment, can significantly alter the treatment plan (Table 48-2). Management always involves vigorous general support of the patient including resuscitation, broad-spectrum antibiotics, and nutritional optimization. The general principles of treatment are to control or eliminate the esophageal leak, maintain or reestablish gastrointestinal tract continuity, eliminate infection and/or obstruction (if any), and drain or debride collections and devitalized tissue.

Cervical perforations usually are best managed by immediate drainage of the retroesophageal space. This procedure is performed via an oblique neck incision along the anterior border of the sternocleidomastoid muscle. The carotid sheath contents are retracted laterally and the visceral compartment medially. A left-sided approach is preferred to reduce the risk of recurrent nerve injury. Instrumental perforations are typically located directly posterior and are found easily. The cavity of contrast material, saliva, and infected contents should be irrigated and drained. Additional tissue planes should not be opened if they are undisturbed. There is no need to search for the offending mucosal tear because the perforation will heal with proper drainage, antibiotics, and nutritional support. Generally, a Penrose drain is placed, supplemented with active suction if dependent mediastinal tracking of infection is present. The patient is kept nothing by mouth for a period of 5 to 7 days depending on clinical status. Drains may be removed either in the hospital or an outpatient setting.

In thoracic perforations, conservative management is considered in two situations: (1) a contained perforation, demonstrated by contrast study, which has already undergone a period of observation with no ill effects and (2) an intramural perforation between the mucosa and muscularis. Cameron et al.⁷ suggested criteria for nonoperative treatment of esophageal perforation: (1) a contained perforation, (2) ready drainage back into the lumen, and (3) minimal symptoms. This series only reported on eight patients, five of which had an esophagogastric leak after resection, thus invalidating the series with regard to treatment of true esophageal perforations. Only two patients suffered from barogenic trauma. All were managed nonoperatively after a period of observation, such that the patients were self-selected. The decision to provide nonoperative treatment is more compelling when there has already been a trial, so to speak, of supportive care only that has been successful. Far more difficult is the decision to pursue nonoperative treatment immediately after diagnosis. Altorjay et al.⁸ reported on 15 transmural perforations (they also reported 6 cases of intramural perforations) treated nonoperatively. In all, 70% were early perforations, and 50% were cervical. Four patients deteriorated and required operation; 2 of 14 patients died (14% mortality). These authors essentially endorsed Cameron's criteria but also added additional recommendations: (1) circumscribed perforation; (2) contrast material flows back into lumen; (3) no cancer, obstruction, or abdominal esophageal leak; and (4) minimal signs and symptoms.

For most perforations of the thoracic esophagus that are diagnosed early (<24 hours), there is general agreement that operative repair is best.^9-12 A thoracotomy is performed on the side of the perforation at the expected interspace of the tear. The pleural space is thoroughly cleansed and the lung decorticated. The esophageal laceration is debrided in conservative fashion, the mucosa is meticulously closed with interrupted sutures, and the muscularis is similarly closed as a second layer¹² (Fig. 48-9). An alternative closure technique is to use a linear stapling device to close the mucosal tear¹¹ (Fig. 48-10). This tear often extends beyond that of the muscularis such that the surgeon must ensure the two ends are well exposed. (This may necessitate extending the muscle tear by incising it to fully expose the underlying mucosal tear.)

Buttressing, even of a fresh perforation, with healthy vascularized tissue of good consistency is advisable. The repair should be checked with instillation of a large volume of methylene blue-tinted saline injected into the esophagus via the nasogastric tube after completion of the suture lines. An optimal buttress is a pedicled intercostal muscle flap, preferably elevated at the time of entry into the chest, such that the vasculature to the muscle is intact. It should be sutured down as if it were being anastomosed to the esophageal wall rather than simply “tacked” over the closure. Other flaps that have been used include the pericardial fat pad, diaphragm, chest wall muscle, omentum, and stomach wall folded over a low perforation. The pleura is not thickened sufficiently by inflammation in early perforation to be used effectively as a flap, although this has been mistakenly advised. Intercostal muscle has been used successfully as a primary patch, rather than as a buttress, in patients with rupture after balloon dilation for achalasia, in whom myotomy could not be employed and esophageal closure was impossible. The pleural space is well drained, and large suction tubes placed immediately adjacent to the repair site is recommended.

Drains should be left in place until a postoperative contrast swallow is completed. Gastrostomy and jejunostomy are advisable, the first to keep the stomach empty and prevent reflux into the esophagus, and the latter for long-term feeding. Any esophageal or pyloric obstruction must be relieved at operation or reperforation is more likely to occur.

In the case of an achalasia patient who is perforated during dilation, the closure should be accomplished in the usual two layers and then a myotomy performed on the opposite side of the esophagus. The standard perforation repair patient is kept nothing by mouth for at least a week, after which a contrast swallow is performed to ascertain whether the repair has been successful. If there is a small, contained leak after repair that is asymptomatic, continued observation and maintenance of the nothing by mouth status is advisable (Fig. 48-11).

Resection for esophageal perforation generally has been restricted to presentation involving a carcinoma, sometimes as a result of biopsy or dilation, or in rare cases of extensive necrosis of the esophagus from infection.^13,14 Some authors have proposed this technique as primary treatment for all intrathoracic perforations. This seems inadvisable because no esophageal substitute ever functions as well as the native esophagus. We believe that late perforations also should be treated surgically in almost every case. Surgery may be vastly more difficult because of inflammation, contamination, and induration of localized tissues. However, the mucosa usually can be closed after limited debridement of its edges. Closure of the indurated muscular wall provides, at best, provisional closure for the purpose of accomplishing the surgery. Definitive sealing of the esophagus from the pleural space is accomplished by the use of a sturdy, well-vascularized tissue buttress applied as described previously. Full expansion of the lung by decortication and cleansing of the pleural space is also essential. In a particularly difficult situation, the diaphragm may be opened and omentum brought up as an additional buttress. The risk of transdiaphragmatic sepsis appears to be small in these cases. Another technique to manage late perforation involves insertion of a large T-tube in the esophagus with the sidearm brought out through the chest wall. This is rarely necessary, and results are not predictable. Esophageal diversion is almost never desirable. This allegedly reversible procedure often prevents salvage of a functional native esophagus, which is far superior to any eventual replacement. Diversion was initially a recommended option because of the high incidence of releakage after primary repair, especially in late cases.¹⁵ Use of primary repair with meticulous buttressing obviates the need for diversion, except in a rare case of extensive esophageal destruction or necrosis. In such cases, esophagectomy with delayed reconstruction is the best option. Continued irrigation per esophagus into the pleural space has been proposed to avoid operative repair.¹⁶ This seems a dubious choice.

The overall mortality from a recent collected case series from 1990 to 2003 was 18%.³ The etiology of the perforation affected outcome. Barogenic perforations had a mortality of 36%, instrumental perforations 19%, and traumatic perforations 7%. The location of the perforation also affected outcome. Cervical perforations had a mortality of 6%, whereas thoracic perforations had a mortality of 27%. Delay in treatment affected outcome, as would be expected. When treatment was initiated within the first 24 hours, average mortality was 14%, whereas it was 27% if treatment was delayed beyond 24 hours. Finally, the type of treatment influenced outcome. Primary repair had a mortality of 12%, resection 17%, drainage 36%, exclusion 24%, and nonoperative management 17%. These results have not changed appreciably since the last large case series review in 1992.¹

Older series suggested a benefit to using a tissue buttress, with reduced leak rates and mortality after primary repair.¹ More recent series have reported good results without use of a tissue buttress.¹¹ Our feeling is that adding a tissue buttress may help contain a small postoperative leak and that the morbidity of taking an intercostal muscle for a flap is minimal. Of greater importance is complete exposure of the mucosal tear with meticulous and secure closure of both the mucosa and muscularis. Older series also suggested that primary repair should be undertaken only for early perforations and that exclusion or resection be the mainstay of management for late perforations. Again, recent reports suggest that primary repair is applicable whether performed early or late. As expected, results are better with early repair, although most patients with late repairs also did well.

Despite advancements in diagnostic imaging, surgical technique, and postoperative care, open surgical repair for esophageal perforation continues to be associated with significant morbidity and mortality, particularly in cases of delayed presentation (>24 hours). Nonoperative management of esophageal perforation was classically reserved for poor surgical candidates due to associated comorbidities precluding surgical repair, or as palliation for perforations associated with malignant disease. The introduction of expandable bare-metal stents changed the management paradigm in this patient population; protocols of supportive care including nothing per mouth, drainage, broad-spectrum antibiotics, and total parenteral nutrition matured to include expeditious stent placement, early resumption of oral intake, and timely hospital discharge. These initial series were difficult to interpret; they included heterogeneous patient populations, a variety of stent types, and varying management protocols, although they uniformly demonstrated good success rates (60%–90%). Increased experience with stents led to expanded use in patients presenting with iatrogenic perforations, particularly after delayed presentation and in those with significant associated mediastinal/pleural contamination. Although their initial use was problematic due to difficulties associated with removal of uncovered metal stents (mucosal ingrowth), several small prospective series demonstrated the utility of stents as a primary treatment option for these patients. This indication has been expanded to included patients following early presentation who traditionally have reasonable outcomes with open surgical repair. The introduction of covered removable self-expanding stent technology, which are easily deployed, restrict mucosal ingrowth, and are adjustable if migration does occur, have led to widespread use by thoracic surgeons for both iatrogenic and benign esophageal perforations.

Although stent deployment for esophageal perforation is a less invasive model, appropriate use mandates equivalent management goals performed during open repair. Namely, rapid closure of the perforation, drainage of associated infection within the pleural space or mediastinum, and enteral access for nutrition during recovery are a necessity. Optimally, stent placement would minimize morbidity and mortality associated with thoracotomy and esophageal repair.

Freeman et al. managed 17 patients following iatrogenic intrathoracic esophageal perforation with silicone stent placement as initial therapy concurrently with mediastinal/pleural drainage by VATS or tube thoracostomy (in cases of delayed presentation or evidence of extraesophageal contamination). Leak occlusion was successful in 94% of patients, and 82% were able to initiate oral intake within 72 hours of stent placement. Only one patient was ineffectively controlled by stent placement and required operative repair. Stent migration required replacement or repositioning in three patients, and all stents were removed within 2 months without stent-related complications. Interestingly, this study included a significant number of sick patients; 65% displayed symptoms of mediastinitis, and 24% of sepsis. Average time from perforation to stent placement was 39 hours. These results are certainly impressive considering the delayed presentation of most patients and evidence of mediastinal contamination. Follow-up endoscopy revealed no evidence of postoperative stricture, likely due to the intrinsic properties of the stent that acts as a rapid manifold for tissue healing.¹⁷

In a subsequent assessment by Freeman et al. of 19 patients with spontaneous esophageal perforations (Boerhaave syndrome), also managed with silicone stent placement, leak occlusion was successful in 89% of patients, and 79% of patients initiated PO intake within 72 hours of stent placement. Mean time for onset of symptoms to stent placement was 22 hours. This study included patients without significant underlying medical problems but high rates of mediastinitis (68%), sepsis (16%), and obesity; contributing factors that increase risk for open surgical morbidity and mortality. Two patients with leaks at the gastroesophageal (GE) junction failed stent management and required operative repair. Four patients required repositioning or stent replacement due to migration. This migration rate was higher than Freeman's previous study, although not unexpected due to stent placement in proximity to the GE junction. Stent removal was individualized based on cause of perforation, anatomic location, patient's nutritional status, and resolution of any associated infection, and was completed in all patients within 1 month of placement with an overall success rate of 79%. Again, no postoperative strictures were reported. These results were surprisingly good considering the difficulty associated with placing/maintaining stent position in proximity to the GE junction.¹⁸

In a similar fashion, D'Cunha et al. managed 15 patients with silicone stent placement in lieu of surgical management for esophageal perforation. Almost 60% of patients were stented within 24 hours of diagnosis, and 45% of patients underwent drainage procedures for source control. Overall success was 60% with mean 27 days to resolution. Median time to return of PO intake was 3 days. All stents were removed without complication and no postoperative strictures were noted. Stent migration occurred in 13% of patients. Most notably an 88% success rate was achieved in their final 17 patients as the author's experience matured. In their final nine cases, all patients were successfully managed with stent placement after diagnosis, with resumption of PO intake by postoperative day 1.¹⁹

Fisher et al. managed 15 patients with spontaneous and iatrogenic esophageal perforations, placing self-expandable metal stents in both groups. Patients were stratified into two groups based on time to intervention: (1) those stented within 45 minutes of perforation (iatrogenic following endoscopic procedures) or (2) patients stented on average 123 hours following injury (Boerhaave's, laparoscopic fundoplication, esophageal diverticulectomy). Patients in the early treatment cohort were managed successfully with stent placement alone and were discharged by hospital day 5. Only one patient required a drainage procedure for fluid accumulation close to the perforation site. In the second group, postoperative management was complicated in some cases by sepsis and multiorgan failure, including one patient death, with an average length of stay of 44 days. Stent extraction was accomplished in 12 patients between 10 days and 8 weeks postoperatively.²⁰

In our own institution, we treated a 63-year-old gentleman who underwent endoluminal stent grafting of a thoracic aortic aneurysm, complicated by an aortoesophageal fistula 4 weeks postoperatively. The patient underwent exploration and an isolated perforation in the esophagus was repaired primarily; buttressed with an intercostal muscle flap. In addition, latissimus dorsi was mobilized and ultimately interposed between the aneurysm sac and the esophageal repair. Postoperatively, the patient developed evidence of sepsis, renal failure, and respiratory failure requiring intubation. A gastrografin swallow demonstrated contrast extravasation from the esophagus (Fig. 48-12A). The patient was brought back to the operating room for endoscopic evaluation, stent placement, and mediastinal/pleural drainage (Fig. 48-12B). Postoperatively he responded to antibiotics, resumed oral intake, and ultimately was discharged to a rehabilitation facility. Contrast esophagram at 1 week demonstrated esophageal coverage and resolution of leak (Fig. 48-12C).

These studies demonstrated that esophageal stents are a viable primary treatment option for patients presenting immediately after esophageal perforation. Likewise, prolonged time to presentation following esophageal perforation and the presence of sepsis or shock from associated contamination does not preclude the use of esophageal stents as primary management. Drainage of associated contamination by VATS or percutaneous tube thoracostomy is suggested in these cases. The time point at which percutaneous or operative drainage is necessary has not been well defined. Because this is a vital component of therapy and ultimately impacts outcome, its use should be strongly considered in all patients with any evidence of contamination or for those presenting in a delayed fashion. Stent extraction after healing should always be performed due to complications associated with long-term treatment. This has been facilitated by the ease of removal inherent to covered/silicone stents. The potential for stent migration requires postoperative monitoring with chest x-ray to ensure adequate positioning; covered stents are inherently resistant to mucosal ingrowth and allow for repositioning or replacement without significant difficulty.

Mandatory to the successful management of esophageal perforation with stents is the fundamental premise that the goals achieved during open surgery are accomplished in an equivalent fashion: closure of the esophageal perforation, drainage of contamination, broad spectrum antibiotics, and prompt resumption of enteral nutrition to enhance healing. Stricture rates of 10% to 50% historically are noted following operative esophageal repair, necessitating further endoscopic procedures, dilatations, and risk. This is an obvious advantage of stent therapy. Success ultimately depends on a uniform approach that includes appropriate patient selection, experience with stent placement, and appropriate postoperative care. Endoscopy remains a critical component in the management of esophageal perforation: at presentation, to document leak resolution, and for follow-up evaluation.

Esophageal perforations are not a single disease, but separate diseases with different symptoms and management dependent upon the mechanism and location of injury. The authors make this clear. I endorse their observation that the mucosal injury in a cervical esophageal perforation does not have to be specifically closed, but that the perforation needs to be adequately drained. I also appreciate the numerous surgical pearls of wisdom included in the descriptions of the operative repairs. I would point out to the reader that I have personally found either a T-tube repair or esophageal diversion to be helpful for the occasional patient with a late perforation and poor tissue quality. Edematous tissue that cannot adequately hold a stitch can be treated well with either of these two strategies.

This chapter makes a significant contribution in specifically addressing intramural perforations. It also presents a very good discussion of the risks and potential benefits of conservative management for carefully selected patients. It is important to note that esophageal perforation remains a life-threatening disease with double digit 30-day mortality within all subgroups. The discussion of the use of stents is balanced and fair. I agree with the authors that this may dramatically affect the management of perforations in the near future. The observation that stents are less successful with GE junction perforations is an important point, and mimics our experience. We no longer use stents for GE junction perforations but approach them laparoscopically. Also, the need to drain any fluid collection in addition to stent placement cannot be emphasized enough. Finally, the anecdotal report of using a stent to salvage a leak following an open repair suggests one possible strategy for the integration of this new technology to improve the outcome of this highly lethal disease.