Reinforced Primary Repair of Thoracic Esophageal Perforation

Transcription

1 Reinforced Primary Repair of Thoracic Esophageal Perforation Cameron D. Wright, MD, Douglas J. Mathisen, MD, John C. Wain, MD, Ashby C. Moncure, MD, Alan D. Hilgenberg, MD, and Hermes C. Grillo, MD General Thoracic Surgical Unit, Massachusetts General Hospital, and Department of Surgery, Harvard Medical School, Boston, Massachusetts Background. Treatment of esophageal perforation, especially when diagnosed late, remains controversial. Methods. Twenty-eight patients were treated for thoracic esophageal perforation with reinforced primary repair regardless of time of presentation. Results. Fifteen patients were treated early (<24 hours) with no deaths. Two had contained postoperative leaks, which healed. Thirteen were treated late (mean, 5.5 days) with four deaths (3 with healed repairs). Postoperative leaks occurred in 7 patients; of the leaks, 4 healed, 2 became a controlled fistula, and 1 required reoperation. Primary healing with preservation of the native esophagus was achieved in 25 patients (89%). Among the 18 patients without evidence of sepsis preoperatively, post- operative leaks developed in 2 (11%). Ten patients had evidence of sepsis preoperatively, and postoperative leaks developed in 7 (70%). Conclusions. Patients who present with sepsis have an increased risk of postoperative leak and therefore should have the repair buttressed. Overall mortality was 14% and no deaths were due to persistent leaks or mediastinal sepsis. Reinforced primary repair retains the native esophagus and avoids the need for later reconstructive operations. In the absence of a nondilatable stricture or cancer, reinforced primary repair should be performed for most thoracic esophageal perforations, early or late. (Ann Thorac Surg 1995;60:245-9) ontroversy continues about treatment (repair, exclu- C sion/diversion, or resection) of thoracic esophageal perforations, especially when treated late. Perforation of the thoracic esophagus remains a devastating illness with mortality rates approaching 25% despite aggressive modern surgical treatment [1]. Delay in diagnosis and treatment significantly influences outcome with mortality rates approximately doubling if treatment is delayed beyond 24 hours [1]. In the absence of esophageal cancer or distal obstruction many thoracic surgeons have favored primary repair, often with a tissue buttress for early (<24 hours) thoracic esophageal perforations [11. However, some surgeons do not advocate primary repair for late perforations (>24 hours) based on time alone and instead advocate exclusion/diversion or resection [2, 3]. We have favored repair and retention of the native esophagus whenever possible without regard to time of perforation [4]. This obviates the need for a second major operation to restore continuity of the gastrointestinal tract. This report reviews our experience with reinforced primary repair of thoracic esophageal perforations over the past 15 years. Presented at the Thirty-first Annual Meeting of The Society of Thoracic Surgeons, Palm Springs, CA, Jan 30-Feb 1, Address reprint requests to Dr Wright, Massachusetts General Hospital, Warren 1212, Boston, MA by The Society of Thoracic Surgeons Material and Methods Patients From July 1979 to July 1994 we performed reinforced primary repair of a thoracic esophageal perforation in 28 patients on the General Thoracic Surgical Service. Hospital records and office charts were reviewed and the patients' clinical status and course analyzed. During the same period, 13 additional patients were treated for thoracic esophageal perforation by methods other than reinforced primary repair: esophageal resection, 8; exclusion/diversion, 2; nonoperative, 2; and T tube diversion, 1. These patients were not analyzed. Fifteen patients were treated early (<24 hours) and 13 patients were treated late (>24 hours). The average age of the patients was 67 years; 19 were male and 9 were female. The cause of the perforation was postemetic in 14, dilation in 6, postoperative in 3, foreign body impaction in 2, esophageal intubation in 2, and esophagoscopy in 1. Preexisting esophageal disease was present in 13 patients: benign stricture in 10 and achalasia in 3. The average delay from perforation to treatment in the early group was 12 hours (range, 1 to 22 hours) whereas it was 5.5 days (range, 1 to 23 days) in the late group. The causes of delayed diagnosis and treatment appeared to be late referral by another hospital to our institution in 14 patients, a delay in diagnosis in 4, and delayed patient presentation in 3. Sepsis was defined by the presence at presentation of fever, tachycardia, and hypotension requiring volume or inotropic resuscitation. Three patients (20%) had evidence of sepsis preoperatively in the early /95/$ (95)00377-W

2 246 WRIGHT ET AL Ann Thorac Surg PRIMARY REPAIR OF ESOPftAGEAL PERFORATION 1995;60:245-9 Mucosal edge under muscle / Torn muscle ~ ~ ~ded / C mueosal edge ~ ~ / Fig 1. Technique of reinforced primary repair of thoracic esophageal perflm~tions. (A) The mediastinal pleura is opened if necessary to fully expose the perforation site. Ragged. partially necrotic muscle at the border qf the tear hides the full extent of the mucosal injury. (B) Necrotic muscle has been debrided back to fully eypose the mucosal defect and the mucosal edge is trimmed to healthy tissue. (C) Inverting interrupted fine 4-0 sutures are placed and tied so that the knots remain inside. Inward traction on the previously placed suture facilitates proper mucosal inversion. Suturing is begun in each corner and finishes in the middle. (D) A second layer of interrupted fine 4-0 sutures are placed if possible in the muscle layer to cover the mucosal repair. (E) A previously harvested intercostal muscle flap with an intact vascular bundle is swung on its posteriorly based pedicle to cover and buttress the repair. This is sutured over the closure with interrupted mattress sutures in a circumferential fashion to make a third water-tight layer. All sutures are placed first beforc ~ing to facilitate an exact repair. group whereas 7 patients (54%) had sepsis in the late group. Technique of Primary Repair and Buttressing Thoracotomy is performed on the side of the perforation. Pleural debridement and decortication are performed if necessary. This allows complete reexpansion of the lung to eliminate potential spaces and the possibility of empyema. The mediastinum is widely opened and debrided of all necrotic debris and tissue. By eliminating dead space, possible empyema, and localized necrotic tissue the possibility of secondary breakdown of the esophageal repair is minimized. Characteristically, the rent in the muscle of the esophagus is smaller than that of the mucosa so muscle must be opened to expose the full extent of the mucosal tear (Fig 1). Failure to do so will lead to persistent leaks and disruption of the repair. Esophageal muscle is debrided back to healthy tissue. The esophagus then is closed in two layers with fine 4-0 interrupted sutures. Occasionally, necrosis or induration of the muscle prohibits twolayer closure and only the mucosa can be closed. The mucosa, in contrast to the muscular layer, is usually in surprisingly good condition. The mucosal layer is closed with an inverting technique maintaining tension on the previously placed suture to facilitate mucosal inversion. The repair is tested by injection of air in a nasogastric tube with occlusion of the distal esophagus while under saline solution or by methylene blue injection. Recognizing that this repair is performed in a contaminated field, buttressing of the repair is prudent. This provides an extra layer of protection to reduce the risk of leak and contain it if a small leak occurs. Intercostal muscle is used preferentially due to its bulk blood supply, and large arc of rotation. Omentum is useful in complicated situations and is probably the most secure means of closure due to its vascularity and resistance to infection. Pleura should not be used in early perforations due to its lack of substance and vascularity. With time and inflammation pleura can become quite robust and better vascularized; it then can act as a reasonable buttress [4]. Whichever choice is made it should be carefully approximated to the esophageal repair with multiple sutures as if doing an anastomosis. Chest drains are placed in the gutter, next to the repair and along the diaphragm. The chest cavity is thoroughly cleansed before closure. A small laparotomy then is performed and a draining gastrostomy and feeding jejunostomy are performed routinely. Enteral feeding by jejunostomy is begun as soon as bowel function returns. Adequate drainage of the stomach is important to avoid reflux of gastric contents on the repair. Broad-spectrum antibiotics are administered especially directed at mouth anaerobes and directed by culture results. A Gastrografin (Squibb Diagnostic, Princeton, NJ) swallow is obtained routinely on postoperative day 7 to 10 to check for leaks. An oral diet is resumed gradually once integrity of the esophagus is assured. Results Twenty-eight patients were treated with reinforced primary repair. The repair was buttressed in all patients: intercostal muscle was used in 15, pedicled pleural flap in

3 Ann Thorac Surg WRIGHT ET AL ;60:245-9 PRIMARY REPAIR OF ESOPHAGEAL PERFORATION 13, omental flap in 4, stomach in 1, and pericardial fat pad in 1. Six patients had double buttressing of the esophageal repair. Fifteen patients were treated with reinforced primary repair within 24 hours with no deaths. Two of these patients (both with sepsis present preoperatively) had asymptomatic contained postoperative leaks, which healed within 1 week. Major complications occurred in 4 patients: Candida sepsis in 1, small bowel obstruction from the jejunostomy in 1, intraabdominal abscess requiring percutaneous drainage in 1, and respiratory failure along with myocardial infarction, sepsis, and pulmonary embolism in 1. None of these patients had postoperative leaks. The average hospital stay in these 15 patients was 23 days (range, 9 to 51 days). One patient required resection of a recurrent benign stricture 4 years after esophageal repair. All other patients have retained their native esophagus and are able to eat a normal diet. Thirteen patients were treated after 24 hours with four deaths (31%). Postoperative leaks occurred in 7 patients: 4 were asymptomatic, contained, and healed; 2 became a controlled fistula; and 1 required reoperation. Seven patients had sepsis preoperatively, and postoperative esophageal leaks developed in 5 of them. Six patients did not have sepsis preoperatively, and 2 of these had postoperative esophageal leaks. Four patients died: 2 of multiorgan failure, I of stroke, and I of an aortoduodenal fistula after repair of a ruptured aneurysm. Three of these 4 patients died with a healed esophageal repair. Death was unrelated to persistent esophageal leakage or mediastinal sepsis. Esophagectomy ultimately was required in 2 patients for a persistent fistula in 1 and esophageal obstruction in another. This last patient was unique and was the only one who required reoperation. This patient was schizophrenic with psychogenic vomiting and underlying Barrett's esophagus with a mild stricture. His postemetic rupture was repaired and a barium swallow on day 9 was normal. On hospital day 13 just before discharge back to the psychiatric hospital he had another bout of psychogenic vomiting and ruptured his repair. Re-repair was successful but the residual lumen was too small too allow oral intake and therefore esophagectomy was performed. Major complications (other than leaks) occurred in 4 patients: respiratory failure in 3, hemothorax in 1, and pneumonia in 1. Only I of the patients with complications had a postoperative esophageal leak. The average hospital stay in the late patients was 47 days (range, 17 to 122 days). All surviving patients are able to eat a normal diet and have not required later operations or dilations. Overall mortality was 14% in patients treated with reinforced primary repair. Primary healing with preservation of the esophagus was achieved in 25 patients (89%). Patients who present to the surgeon with sepsis preoperatively appear to have an increased risk of leakage after repair: 70% as opposed to only 11% if no evidence of sepsis is present. There were 9 postoperative leaks after buttressed repair; all were asymptomatic, 7 healed with observation only, and only 2 persisted as a fistula. Comment The management of thoracic esophageal perforation remains controversial in large part due to the variability in cause, underlying esophageal disease, treatment delays, and management strategies. Randomization of treatment has not been reported and therefore comparison of results of treatment of similar patients must suffice to judge the merits of a particular treatment approach. The majority of surgeons now agree that reinforced primary repair of an early esophageal perforation or resection of a perforated cancer or nondilatable stricture is an appropriate therapy [1]. Extension of the principle of primary repair of simple (postemetic or instrumental) perforations to those with achalasia or a soft benign stricture has been widely accepted as long as distal obstruction can be relieved [1]. This has been achieved by esophagomyotomy for achalasia or intraoperative dilation for a stricture [5, 6]. Controversy continues about treatment of late perforations. Strong advocates support exclusion/ diversion or resection and advise against primary repair of late perforations [2, 3, 7]. We have had a long interest in primary repair of late perforations and believe our results support this view [4, 8]. Our overall mortality of 14% (0% early, 31% late) is relatively low and compares favorably with other series of perforations regardless of treatment [1]. Deaths after repair have been related to the sepsis syndrome with multiorgan failure (with healed repairs) or unrelated conditions (stroke, complication from ruptured abdominal aortic aneurysm) and are not due to failure of the repair. Complications are not increased in patients with contained postoperative leaks. Complications, death rates, and length of hospital stay all are increased in patients who have delayed treatment, as would be expected. These findings have been reported by others [1]. Early diagnosis and treatment of esophageal perforation remains the most important principle to reduce morbidity and mortality from this illness. Repair of the perforated esophagus without a tissue buttress often fails, leading to a clinical leak rate of about 40% I91. The rate exceeds 50% if the repair is delayed beyond 24 hours [10]. However, if the repair is buttressed, leak rates are reduced dramatically to about 10% [9]. Experimental studies also have documented markedly reduced leak rates after buttressed repair of esophageal perforation ]11]. More importantly, the nature of a possible leak is changed from a potentially fatal free intrapleural leak to an asymptomatic, contained leak that will heal with observation. Seven of our nine leaks healed in this fashion. Only two fistulas developed for a "clinical" leak rate of only 7%. Careful technique, including buttressing the repair as if it were an anastomosis (as opposed to a few tacking sutures) as well as broadspectrum antibiotics and aggressive enteral nutritional support contributed to these good results. Preoperative sepsis may be a risk factor for leak after repair and probably is a marker of severity of disease. The presence of preoperative sepsis mandates a buttressed repair to contain a leak. Primary healing with preservation of the

4 248 WRIGHT ET AL Ann Thorac Surg PRIMARY REPAIR OF ESOPHAGEAL PERFORATION 1995;60:245-9 esophagus was achieved in 25 patients (89%). Esophagectomy is rarely required in the absence of cancer or a nondilatable stricture. Selection of treatment for thoracic esophageal perforation must remain individualized and take into account many variables including cause, duration, underlying esophageal disease, and condition of the patient. In the absence of cancer or an undilatable stricture, reinforced primary repair is the preferred treatment in most patients [1, 4, 8, 12, 13]. With the principles outlined, success can be achieved in the majority of patients. Surgeons should be familiar with other options of treatment (esophagectomy, exclusion/diversion, T tube drainage, nonoperative) to allow individualization of treatment. References 1. Jones WG, Ginsberg RJ. Esophageal perforation: a continuing challenge. Ann Thorac Surg 1992;53: Flynn AE, Verrier ED, Way LW, et al. Esophageal perforation. Arch Surg 1989;124: Salo JA, Isolauri JO, Heikkila LJ, et al. Management of delayed esophageal perforation with mediastinal sepsis. Esophagectomy or primary repair. J Thorac Cardiovasc Surg 1993; 106: Grillo HC, Wilkins EW. Esophageal repair following late diagnosis of intrathoracic perforation. Ann Thorac Surg 1975;20: Miller RE, Tiszenkel HI. Esophageal perforation due to pneumatic dilation for achalasia. Surg Gynecol Obstet 1988; 166: Orringer MB, Stirling MC. Esophagectomy for esophageal disruption. Ann Thorac Surg 1990;49: Urschel HC, Razzuk MA, Wood RE, et al. Improved management of esophageal perforation: exclusion and diversion in continuity. Ann Surg 1974;179: Michel L, Grillo HC, Malt ILA. Operative and nonoperative management of esophageal perforation. Ann Surg 1981;194: Gouge TH, Depan HJ, Spencer FC. Experience with the Grillo pleural flap procedure in 18 patients with perforation of the thoracic esophagus. Ann Surg 1989;209: Nesbitt JC, Sawyers JL. Surgical management of esophageal perforation. Ann Surg 1987;53: Bryant LR. Experimental evaluation of intercostal pedicle grafts in esophageal repair. J Thorac Cardiovasc Surg 1965; 50: Ohri SK, Liakakus TA, Pathi V, Townsend ER, Fountain SW. Primary repair of iatrogenic thoracic esophageal perforation and Boerhaave's syndrome. Ann Thorac Surg 1993;55: Whyte RI, Iannettoni MD, Orringer MB. Intrathoracic esophageal perforation. The merit of primary repair. J Thorac Cardiovasc Surg 1995;109: DISCUSSION DR DAVID J. SUGARBAKER (Boston, MA): I would like to ask Dr Wright a question, and comment first on a very nicely presented paper and nice discussion of this very devastating problem to the patient and many times to the entire staff caring for the patient. In that late group of patients there was a 70% leak rate, four deaths, 2 patients who required subsequent esophagectomy, and an average length of stay in excess of 40 days; given all of those factors, is there another alternative approach to this one that you would consider or that you would suggest at this time should be looked into, or are you content with those figures? DR RICHARD I. WHYTE (Ann Arbor, M[): Doctor Wright, 1 enjoyed your presentation a lot. It was very, well done. As you know, we also espouse primary repair whenever possible, but we do not buttress the repair. I was wondering what the evidence is that buttressing is necessary or even helpful. Second, you make the assumption that the functional result after primary repair is clearly worse than esophagectomy. I do not know if there is a lot of data to support that contention. Most of the published reports on esophageal repair or perforation deal with the acute episode in the early recovery, and there are really very few data dealing with the long-term results of this condition. DR JAMES W. PATE (Memphis, TN): 1 also congratulate Dr Wright and associates on their excellent clinical results and for emphasizing two important principles: the benefit of viable, healthy muscle buttresses over esophageal suture lines and the fact that it is never too late to repair and salvage a perforated esophagus. We have had repairs as late as 49 days after perforation with salvage of a functioning esophagus [1]. Whatever delay alone contributes, if anything, to prognosis and mortality, I think the cause of the perforation is more important. I have taken the liberty of rearranging some of Dr Wright's data. Half of his cases (14) were postemetic perforations (Boerhaave's syndrome); iatrogenic perforations during dilation was next. These are quite different, etiologically. In the late cases, there were no instrumental perforations; they were all Boerhaave's disease. Of the Boerhaave's cases, 62% had a delay in diagnosis; 63% had preoperative sepsis with shock, among whom 3 (60%) had leaks and 2 (66%) died. All iatrogenic perforations were promptly diagnosed and operated on with minimal sepsis and postoperative leaks and no deaths. The different results can be attributed to the different preoperative status, related to cause, rather than to just time. Thus, the cause appears to be the basic determining factor in outcome. DR SAFUH ATTAR (Baltimore, MD): My question relates to the 30% mortality, in your late cases. This is an acceptable mortality considering the high rate of mortali~ in these patients. However, in retrospect, would you think that a different operation, as Dr Sugarbaker asked, such as esophagectomy would have been better than the complex procedures you went through? In our previous series we had 8 cases treated by esophagectomy due to either perforated cancer or to ulcerations from caustic material and the survival rate was much better; we had much less than 30% mortality. DR STEVEN J. MENTZER (Boston, MA): Doctor Wright, I enjoyed your excellent paper. I noted that you have treated at least one esophageal perforation with a T tube. Given the amount of morbidity related to prolonged sepsis, I wondered if you might comment on the use of a T tube or alternative drainage techniques should a leak occur after primary repair. DR S. K. MOHANTY (Huntington Beach, CA): What I want to ask is what kind of suture you use for your primary repair, not

5 Ann Thorac Surg WRIGHT ET AL ;60:245-9 PRIMARY REPAIR OF ESOPHAGEAL PERFORATION the buttressing as much. Also, how often in late cases were you able to do a two-stage repair? Finally, would you comment on the data published by Dr Skinner, if I am not mistaken, about esophagectomy in late cases? DR JOSEPH E. BAVARIA (Philadelphia, PA): I thought this was a wonderful presentation. My question is, would you change or alter your approach to esophageal perforation in a trauma case, which we have actually seen quite a bit these days in the gunshot wound population. Would you use muscle flaps all the time? DR WRIGHT: First of all, obviously the reason why we are still talking about various treatment options is that there is no randomized study. I doubt that one will ever be done. We do have strong opinions representing the merits of primary repair and I think there are results that substantiate these opinions. To address Dr Sugarbaker, I pointed out that we thought, after careful review of these patients, that the deaths and complications were unrelated to leaks. The length of stay and the postoperative problems that these patients had were related to their debility and sepsis that were present preoperatively. To address Dr Whyte, we buttress the repair because it is a belts-and-suspenders approach. We believe that when you are operating in a contaminated field the repair often will leak. If the repair is buttressed we have changed a devastating postoperative complication into a minor nuisance. Seven of our nine postoperative leaks were contained and healed with no further treatment. So we think ave D' important component of primary repair is to buttress these repairs. With regard to the functional results of primary repair, we did not perform esophageal function studies on these patients postoperatively. We followed them up for a minimum of I year. One patient 4 years postoperatively had development of a recurrent benign stricture, which was nondilatable, and eventually needed an esophagectomy for that. No other late patients had esophagectomies. All other patients have been able to eat a normal diet. To address Dr Pate, I also looked at the causes of the perforations and compared the death rates. For postemetic perforation, the overall death rate was 14% with no deaths in 6 early patients and 2 deaths (25% mortality) in the late group. Accordingly, we think that late presentation is a risk factor, but nonetheless a repair still can be performed, To address Dr Attar's question, I would just reemphasize what [ mentioned to Dr Sugarbaker, that the postoperative complications that these patients get are because of prolonged preoperative mediastinitis and sepsis and do not relate to what kind of operation you do on them. We do perform esophagectomy for perforated carcinoma and perforated nondilatable stricture, and we think it is a very good operation for those two indications. If you have failed primary repair in a late septic patient if would be quite reasonable to perform a resection. To address Dr Mentzer, we did treat one of our 41 patients with a T tube. This was a massively obese lady with bilateral empyemas, adult respiratory distress syndrome, renal failure, and respiratory failure. She had a 30-day-old well-contained perforation and we were able to put a T tube in her to control her leak and connect it to a fistula. However, most perforations even when diagnosed late can be closed successfully with a buttressed primary repair. Reference 1. Pate JYV, Walker WA, Cole FH Jr. Spontaneous rupture of the esophagus: a 30-year experience. Ann Thorac Surg 1989;47: