Kaplan-Meier survival curves of stomach and colon groups of patients. In the stomach group, median survival was 12.8 months; 5-year survival, 19.5%. In the colon group, median survival was 10.4 months; 5-year survival, 15.1% (P = .40). Groups are described in the "Methods" section.
Davis PA, Law S, Wong J. Colonic Interposition After Esophagectomy for Cancer. Arch Surg. 2003;138(3):303-308. doi:10.1001/archsurg.138.3.303
Copyright 2003 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2003
The use of colonic interposition in esophageal replacement after esophagectomy for cancer results in similar morbidity, mortality, and long-term outcome compared with gastric transposition.
Prospectively collected database on patients with esophageal cancer from January 1, 1982, through December 31, 2000.
Academic university hospital department of surgery.
We compared 42 patients who underwent colonic interposition (colon group) with 959 patients who underwent gastric transposition (stomach group) after esophagectomy.
Main Outcome Measures
Morbidity, mortality, and long-term survival.
Greater blood loss (median, 1000 vs 700 mL; P<.001) and longer operation duration (median, 270 vs 225 minutes; P<.001) were encountered in the colon group. We found no difference in cardiopulmonary complications, but we found significantly greater incidences of anastomotic leakage (14.3% vs 3.9%; P = .007) and intra-abdominal septic complications (9.5% vs 0.2%; P<.001) in the colon group. Conduit ischemia developed in 5 patients (0.5%) in the stomach group, 3 of whom underwent successful staged reconstruction with colon. One patient (2.4%) in the colon group was found to have conduit ischemia and died. Hospital mortality rates included 7 patients (16.7%) from the colon group and 102 (10.6%) from the stomach group (P = .21). These figures improved to 0 and 27 (5.5%), respectively, in the second half of the study period (P>.99). Median survival was 12.8 and 10.4 months in the stomach and colon groups, respectively (P = .4).
Colonic interposition is a more complex procedure with increased morbidity, compared with gastric transposition. Overall mortality and survival, however, were similar to those for gastric transposition.
VARIOUS OPTIONS are available for reconstruction of the esophagus after esophagectomy, namely, the choice of conduit (stomach, colon, or jejunum), route of reconstruction (orthotopic, left or right chest, retrosternal, or subcutaneous), and site of the anastomosis (thoracic or cervical). The stomach and colon, as opposed to the jejunum, can be easily transposed to the neck. Preparation of a long jejunal loop is also cumbersome. The jejunum is mostly used to replace the stomach when total gastrectomy is performed, or when a longer loop is used to replace the distal esophagus after esophagogastrectomy. Although the method of reconstruction has no apparent impact on oncological resection, it may affect operative morbidity and mortality and long-term quality of life.
The colon has reportedly been used for esophageal bypass since 1911.1- 3 In 1969, a review reported that "the colon has been the most commonly used visceral substitute since 1950."4 The stomach tube, however, has become increasingly popular. It has many advantages. It is easy to prepare, its vascular supply is robust, and its length is generally adequate, even when brought up to the neck. It has become widely accepted as the first choice for esophageal replacement.5
There are instances when the stomach cannot be used, such as the presence of previous gastric resection or if tumor involvement of a substantial part of the stomach dictates its removal. In these situations, use of the colon is preferred. Proponents believe that the advantages of using the colon outweigh its disadvantages, such as a more complex procedure and hence the potential for increased risk for complications. For most surgeons, however, colonic interposition remains an infrequently performed procedure.6 It is arguable whether this type of complicated reconstructive technique should be limited to dedicated institutions; recent studies have shown that good results are mostly obtained in high-volume centers.7,8
The aim of this study was to review our experience in the use of colonic interposition for esophageal cancer during an 18-year period. We studied its indications, morbidity and mortality rates, and long-term results.
Patients with cancer of the esophagus or gastric cardia treated at the Department of Surgery, University of Hong Kong Medical Centre, Queen Mary Hospital, Hong Kong, were entered into a prospective esophageal cancer database. Patients who underwent esophagectomy and reconstruction using a colonic interposition were identified. The hospital medical charts were obtained and reviewed when additional information was required. This group of patients was compared with those in whom the whole or the distal stomach was used for reconstruction. All patients had a cervical or intrathoracic anastomosis. Patients who underwent proximal gastrectomy for cardia or proximal gastric cancer and who underwent reconstruction in the abdomen only were excluded from analysis.
The choice of surgical approach depended on the location and extent of the tumor, and on results of the cardiovascular assessment of the patient. The various techniques have been described in previous publications from our group.9- 11 In brief, transthoracic resection via a right-sided thoracotomy was preferred. In patients with compromised cardiopulmonary functions, a transhiatal esophagectomy was performed. When the stomach was used for reconstruction for middle- and lower-third tumors, the esophagogastrostomy was usually placed in the right thoracic cavity. In those with tumors of the superior mediastinal segment, a 3-phase esophagectomy was performed with reconstruction to the neck.
In patients who had colonic interposition, the right colon based on the middle (with or without the right) colic vessels was more often used. The terminal ileum was mobilized, together with the ascending and transverse colon, and was brought up to the neck with the terminal ileum anastomosed to the cervical esophagus. When the left colon was used, it was based on the ascending branch of the left colic vessel, with the transverse colon being anastomosed to the cervical esophagus in an isoperistaltic fashion. The primary choice of conduit in the series was the stomach. Colonic interposition was performed when the use of stomach was not possible.
The preferred method of anastomosis to the esophagus, regardless of the conduit used, was a single-layer continuous technique using a fine 4-0 monofilament suture. The technique has been described previously.9 The circular stapler was used in the early part of the series, and it was compared with the hand-sewn technique in a prospective, randomized trial.12 Most anastomoses performed in the neck were hand sewn (97%), compared with 56% of intrathoracic anastomoses. In recent years, all anastomoses, regardless of location, were hand sewn.
Cardiovascular complications included any arrhythmia, myocardial infarction, heart failure, and pulmonary embolus. Major pulmonary complications were defined as the development of aspiration pneumonia, bronchopneumonia, respiratory failure, and acute respiratory distress syndrome. Other medical complications included stroke, renal failure, liver failure, and multiple-organ failure. Deaths within 30 days after surgery and hospital mortality rates within the same hospital admission were reported. Patients were followed up monthly for the first year and every 3 months afterward. At follow-up, swallowing function was assessed as good in those patients with no reported sensation of dysphagia and who tolerated a normal rice diet.
The tumor location in the esophagus, histological type, and disease stage were determined according to the International Union Against Cancer classification.13 Continuous variables are expressed as mean (SD) when the variable is parametric and as median (range) when it is nonparametric. We calculated survival with the Kaplan-Meier method, and we compared the difference with the log-rank test. Statistical significance was taken at P<.05. All calculations were performed with SPSS statistical software (version 9.0; SPSS Inc, Chicago, Ill).
From January 1, 1982, through December 31, 2000, 1801 patients with cancer of the esophagus or gastric cardia were referred for primary assessment and treatment at the Department of Surgery. Of these, 1131 underwent esophagectomy (resection rate, 63%). Esophagectomy was performed with reconstruction using stomach in 1047 patients, jejunum in 42, and colon in 42 (colon group). The whole or distal stomach was used as the conduit with anastomosis in the neck or thorax in 959 patients (stomach group). In the colon group, the reasons for not using stomach as the conduit for reconstruction were previous gastrectomy in 20 (48%), stomach involvement by tumor in 17 (40%), peptic ulcer in 3 (7%), and other technical reasons in 2 (5%).
The demographic features of the 2 groups are shown in Table 1. In the colon group, previous Billroth I and II gastrectomy had been performed in 4 and 16 patients, respectively. In the stomach group, 20 (2.1%) of the 959 patients had undergone previous gastroduodenal surgical procedures for benign diseases such as omental patch repair of perforated peptic ulcer, vagotomy, and pyloroplasty. Those who underwent colon interposition had a higher prevalence of adenocarcinoma, because of more frequent tumor involvement of the stomach necessitating gastrectomy. Barrett adenocarcinoma of the lower esophagus is very uncommon in Hong Kong.
The surgical approaches and route of reconstruction are shown in Table 2. In the colon group, the right colon was used in 28 patients (66.7%), left colon in 12 (28.6%), and transverse colon in 2 (4.8%). In the stomach group, the whole stomach was used in 778 patients (81.1%) and distal stomach in 181 (18.9%). The operation time in the colon group (median, 270 minutes; range, 160-465 minutes) was longer than that in the stomach group (median, 225 minutes; range, 90-520 minutes) (P<.001). Blood loss was also greater in the colon group (median, 1000 mL; range, 300-2500 mL) than in the stomach group (median, 700 mL; range, 100-7000 mL) (P<.001).
The stage distribution of tumors is shown in Table 3, with no significant differences between the colon and stomach groups.
The incidence of medical and surgical complications is shown in Table 4. A greater proportion of patients in the colon group developed anastomotic leaks. Of these 6 patients (14.3%), 5 had leaks from the esophagocolic and 1 from the intestinal anastomosis. Three of these patients died in the hospital. In the stomach group, a leak from the esophagogastric anastomosis developed in 37 patients (3.9%), 14 of whom died in the hospital.
In the colon group, the death of 1 patient (2.4%) 2 weeks postoperatively was due to respiratory failure and sepsis. At subsequent postmortem examination, hemorrhagic ischemic necrosis of the colon was found to have developed. In the stomach group, conduit gangrene presenting as leakage and sepsis on days 4 to 10 developed in 5 patients (0.5%). In 3 patients, this was managed by resection of the necrotic part of the stomach, cervical esophagostomy, and feeding gastrostomy. After 2 months, staged reconstruction using a retrosternal colonic interposition was performed, and recovery was smooth thereafter. In 1 patient, the ischemic fundus was excised, and primary closure was performed. In a second patient with gastric ischemia due to torsion of a redundant stomach, after excision of the necrotic part, the stomach was tunneled subcutaneously and reanastomosed to the cervical esophagus. Both patients died in the early postoperative period.
The 30-day mortality rate for the stomach group was 26 patients (2.7%); for the colon group, 2 (4.8%) (P = .33). Hospital mortality rates were 102 patients (10.6%) for the stomach group and 7 (16.7%) for the colon group (P = .21). Surgical results improved over time. From 1982 to 1990, 75 (15.9%) of 472 patients in the stomach group died in the hospital, compared with 27 (5.5%) of 487 during 1991 to 2000. The respective figures for the colon group were 7 (24%) of 29 and 0 of 13 patients.
During follow-up, 22 patients in the colon group (52.4%) reported good swallowing function and were able to tolerate a normal diet. Eight patients (19.0%) experienced a hold-up sensation, but investigations showed no stricture. An additional 8 (19.0%) were found to have a stricture, located at the upper anastomosis in most cases, that required therapy. In 1 patient, it involved a narrow segment in the midportion of the ileum that was brought up to the neck for anastomosis. These were all treated by means of dilation at endoscopy. In the patient with ileal stricture, after multiple attempts at endoscopic dilation, the stricture was stented by means of a self-expanding metallic stent when locoregional disease recurrence was proven. By comparison, benign anastomotic stricture developed in 193 patients (20.1%) in the stomach group, similar to the incidence in the colon group.
Bowel function was generally good in patients with colonic interposition. Only 3 patients (7%) reported protracted symptoms of diarrhea up to 3 times per day. One patient underwent reoperation at 2 months after surgery to refashion the anastomosis because of a segment of redundant ileum in the neck. Kaplan-Meier survival curves of the 2 groups are shown in Figure 1, with no difference in long-term survival.
Esophagectomy for cancer has the following 2 main aims: cure of the disease and relief of dysphagia. The clinical decision making in the treatment of esophageal cancer is a process of balancing the risks of a particular treatment against the potential benefits gained in survival and quality of life. The choice of conduit of reconstruction may have a significant impact on these considerations.
An ideal conduit should have an adequate length to reach the cervical esophagus and a reliable blood supply and should allow good function for swallowing. The surgical procedure should have a low risk for complications. Proposed advantages and disadvantages of stomach and colon conduits by many investigators are outlined in Table 5.14- 20
The stomach is generally preferred in esophageal replacement because of its simplicity in preparation and reliability compared with colonic interposition. It has been our practice to use the colon for reconstruction only in patients whose stomach had been resected or was diseased. The jejunum, whether as a Roux-en-Y loop, an interposition between the esophagus and duodenum, or fashioned into a pouch, is of particular value after resection of tumors of the gastric cardia or lower esophagus.21,22 For tumors higher in the esophagus when a long loop is required to be placed in the chest or neck, preparation of the jejunal loop is tedious and more difficult and is generally not preferred.9 In reconstruction after extirpation of hypopharyngeal tumors, however, a free jejunal graft with a microvascular anastomosis is excellent, for it avoids a more extensive pharyngolaryngeal esophagectomy using a gastric pull-up.23 A jejunal loop is also useful when the stomach and colon have failed.
Early outcomes of operative mortality and morbidity, such as anastomotic leaks, conduit gangrene, and cardiopulmonary complications, have been prominent in the debate of which conduit is best. Major reports in the literature of colonic interposition as an esophageal substitute are summarized in Table 6.14- 20 In specialized centers, use of colonic interposition is no doubt safe. In our patients, blood loss was greater and operating time was longer in the colon group, reflecting the complexity of the operation. Surgical problems such as anastomotic leakage and septic events were also more common. With better perioperative care, complications are better treated,24,25 but this increased morbidity has to be taken into consideration, especially when these procedures are performed in units with less experience in managing complications after esophagectomy. With increased experience, mortality for both groups was markedly reduced in the second half of the study period. The decline in the hospital death rate after esophagectomy in recent years has been largely attributed to factors that reduce pulmonary complications, such as epidural analgesia and use of bronchoscopy.8
Anastomotic leakages are usually related to poor technique of anastomosis, tension, and blood supply of the conduit. Tension is not a significant problem, in our experience, as both the stomach and colon conduits can reach the neck with ease. The exact reason why the colon had a higher leakage rate is uncertain, but it is probably related to the relatively poorer blood supply compared with the stomach. It has been suggested that routine preoperative angiography would help to select the most appropriate colonic conduit.26 This has not been our routine practice.
When an anastomotic leak occurred in the neck, treatment usually involved opening the neck wound for drainage to wait for healing to take place. Cervical leaks, however, may not be confined to the neck, and upper mediastinal extension may be present. Placement of drains was sometimes required in the operating room. For intrathoracic anastomotic leaks, effective drainage is essential. Radiologically placed drainage tubes were sometimes helpful, but if sepsis was uncontrolled, open debridement and drainage had to be performed. In this regard, treatment for leakage from esophagocolic or esophagogastric anastomoses did not differ. Our previously published randomized controlled trial comparing the hand-sewn and stapled intrathoracic esophagogastric anastomosis showed that leakage rates were similar, but that long-term stricture rates were much higher using the circular stapler.12 For the colon, the number of patients was too small to make meaningful comparisons.
The left colon is favored by some surgeons,14,17- 19 in part because its blood supply has been shown to be more reliable in anatomic studies.27 We prefer to use the right colon. It has also been used successfully by others,16,17,28 with a low incidence of conduit ischemia. Our incidence of colon ischemia compares well with that of 3% to 9% reported for the left colon.14,15,17- 19 The left colon is also preferred by some because of its smaller diameter compared with the right. It is our experience that when incorporating the terminal ileum that is brought up to the neck for anastomosis with the esophagus, the size of the ileum matches well with that of the esophagus.
Unique to the colonic conduit is the risk for redundancy that may be related to technical failure and that has been reported for 15% to 30% of patients. This can also manifest years later.29 Redundancy can cause obstructive symptoms such as dysphagia and regurgitation, and correction can be a complex undertaking. Very few cases of revision are reported in the literature, and among our patients, only 1 required a revision of a redundant loop in the neck. There is no reliable method that we know of that can prevent such complications from taking place.
A colon conduit has been suggested to be more durable, and the supposed long-term functional benefits of colon interposition make it the preferred esophageal substitute in those with benign disease and in patients whose cancer stage predicts long-term survival.14 A colonic conduit provides good long-term swallowing function, and normal oral intake is restored in 65% to 88% of patients with cancer of the esophagus.14,19 Colonic conduits are reported to have active peristalsis, and this is presented as an explanation for their superior function as an esophageal substitute compared with a passive gastric conduit.30,31 Although peristalsis can be demonstrated immediately after surgery,32 long-term emptying likely relies on gravity.33 When the distal stomach is retained in the abdomen after a colon interposition with a cologastric anastomosis, the latter provides additional reservoir function.14
Patients who have an intrathoracic stomach often experience postprandial discomfort and early satiety, probably related to loss of normal gastric function such as receptive relaxation. Patients also experience acid reflux, possible gastric ulceration, and dysfunctional propulsion.17 In addition, Barrett esophagus has been reported to develop in the esophageal remnant. These are important considerations, although in our experience, serious problems are uncommon in long-term survivors after esophagectomy with a gastric conduit. We have never seen a Barrett adenocarcinoma in the esophageal remnant develop in any patient.
Prospective quality-of-life studies are required to properly assess the long-term function of gastric or colonic conduits. For most patients with advanced esophageal cancer, however, performing a safe esophagectomy is of paramount importance, and given the ease of preparation and reliability of the gastric conduit, it will remain the preferred organ for esophageal substitution for most surgeons. Colonic interposition is an essential technique of esophageal reconstruction when the stomach is not available, and is used to salvage those patients with gastric necrosis.
Corresponding author and reprints: John Wong, PhD, FRACS, Department of Surgery, University of Hong Kong Medical Centre, Queen Mary Hospital, 102 Pokfulam Rd, Hong Kong (e-mail: email@example.com).
Accepted for publication November 16, 2002.