Oesophageal cancer is an uncommon and often lethal malignancy. Unfortunately, symptoms rarely occur until late in the clinical course. The goal of surgical treatment remains relief of dysphagia and cure. Combined modality treatment represents a new approach to oesophageal cancer and may offer improvements in survival of selected patients.

There is considerable variation throughout the world in the incidence of squamous cell carcinoma of the oesophagus. The incidence ranges from 5 per 1000000 in the United States to 100 per 100000 in Linxian, China (Table 1) 293. There is also a marked difference in death rates within each country. The peak age of occurrence is between the ages of 50 and 70 years with a median age of death of 66 in the United States. In Western nations, oesophageal cancer is much more common in men than in women (about fivefold). Oesophageal cancer is more common (threefold) in black than in white subjects in the United States. In 1980 squamous cell carcinoma represented 90 per cent of oesophageal cancers and adenocarcinomas only 10 per cent. Recent reports indicate an increase in incidence of adenocarcinoma. In some reports, adenocarcinoma has accounted for more than half of all cancers of the oesophagus. Many of the adenocarcinomas have occurred in Barrett's mucosa, largely in caucasian males.

Many epidemiological studies confirm that heavy alcohol users have an increased risk of oesophageal cancer and that the risk is substantial in very heavy drinkers. When combined with cigarette smoking, the risk is even greater.

Studies confirm that smoking leads to an increased risk of oesophageal cancer also in a dose dependent fashion. Moderate smokers have a fivefold increase in risk. This is true for cigar and pipe smoking as well as for cigarette smoking.

The geographical variability of oesophageal cancer strongly correlates with areas that have poor nutritional status. Nutritional factors beyond malnutrition that have been implicated include soil deficiencies in zinc, molybdenum, magnesium, and iron. The clearest relationship exists with molybdenum deficiency in the soil, which leads to accumulation of nitrates and nitrites in plants in turn leading to nitrosamines, a known oesophageal carcinogen. Plummer–Vincent (Patterson–Kelly) syndrome is associated with carcinoma of the upper third of the oesophagus and is associated not only with iron deficiency anaemia, but also vitamin B deficiencies. Coeliac disease (non-tropical sprue) is associated with an increased risk of oesophageal cancer perhaps due to the malabsorption syndrome that leads to multiple nutritional deficiencies.

Oesophageal disorders associated with an increased incidence of carcinoma include chronic lye strictures, achalasia, and perhaps oesophageal diverticuli. Long-term use of spicy and hot drinks may also predispose to chronic irritation and are implicated in the pathogenesis of oesophageal carcinoma. Tylosis is a rare autosomal dominant disease associated with an increased risk of carcinoma of the oesophagus.

Until recent years, the majority of oesophageal cancers were squamous in histology. For purposes of classification, the oesophagus may be divided into three anatomic areas: the upper, middle, and lower third. The upper third (cervical oesophagus) extends from the cricopharyngeal sphincter to the thoracic inlet, the middle third from the thoracic inlet to 10 cm above the gastro-oesophageal junction and the lower third from 10 cm above the gastro-oesophageal junction to the cardia of the stomach. About 15 per cent of squamous cell cancers occur in the upper one-third, 50 per cent in the middle third, and 35 per cent in the lower third.

Three gross patterns of growth are commonly seen. The ulcerating type has a well-defined ulcer with raised irregular edges. The fungating type has a large intraluminal growth component with an irregular surface. The infiltrating type has extensive intramural growth, often circumferential, with minimal ulceration. There is often a significant submucosal extension of tumour far from the visible edge of mucosal tumour (longitudinal spread), more so in a cephalad fashion than a caudad one. In one study, cancer was present 6 cm cephalad from the primary tumour in 22 per cent of patients and 9 cm from it in 11 per cent of patients. The caudad growth was rarely more than 5 cm distal. These data underlie the importance of wide resection of the oesophagus and frozen section analysis of resection margins to ensure complete removal of the tumour.

Transmural tumour penetration (vertical growth) is present in the majority of cases that come to surgery. Lack of a serosal covering of the oesophagus may explain the early invasion of mediastinal structures. Mediastinal structures commonly invaded include the trachea and left main bronchus, aorta, pericardium, and pleura. A malignant airway–oesophageal fistula occurs more commonly than an aorto-oesophageal fistula.

Lymph node metastases are common and correlate with the depth invasion of the tumour. For lesions limited to the submucosa (T1) there is only a 14 per cent incidence of positive nodes. (For discussion of staging, see below.) The incidence of nodal metastases rises from 30 per cent for lesions limited to the muscularis propria (T2), 50 per cent for lesions up to the adventitia (T3) and 75 per cent when the tumour invades the peri-oesophageal tissues (T4). Lymph node metastases may be found at a distant site from the primary tumour because of the extensive longitudinal spread of tumour via the submucosal lymphatic vessels. The lymphatic drainage of the oesophagus is primarily longitudinal rather than segmental, further complicating the surgical treatment of oesophageal cancer. Akiyama has convincingly demonstrated that lymph nodes are often positive at the opposite end of the oesophagus from the carcinoma regardless of whether the cancer is in the upper or lower oesophagus (Table 2) 294. Visceral metastases involving the lung, liver, bone, kidneys, pleura, and brain are common at death.

The natural history of oesophageal cancer may be inferred on the basis of extensive mass screening studies done in China on high-risk populations. Screening was performed primarily by abrasive cytology utilizing a swallowed balloon that is inflated and pulled back through the oesophagus thus collecting a representative sample of epithelial cells. Dysplastic cells are almost always found in association with cases of early carcinoma. Furthermore, dysplasia can progress to carcinoma although not invariably so. In one study of over 1500 patients followed for up to 12 years, 15 per cent developed carcinoma if severe dysplasia was present. Only 1 per cent developed oesophageal cancer with mild dysplasia and no carcinomas were found in the control group with normal cytologies. In a separate study of 14000 subjects, the average age of those with severe dysplasia was 52, whereas those with carcinoma was 57, suggesting a 5-year lag time in developing carcinoma once severe dysplasia develops. Not all patients with dysplasia will develop carcinoma; in one study 45 per cent of mildly dysplastic and 40 per cent of severely dysplastic cells returned to normal over several years. In patients with early cytologically diagnosed carcinoma who refused treatment and were followed, about one-half remained asymptomatic in the early stage of the disease for a mean period of 75 months. The other one-half developed symptomatic late carcinoma an average of 55 months after diagnosis. Thus, the development of symptomatic carcinoma takes many years and allows for ample time for early diagnosis. Surgical treatment of early, cytologically diagnosed, asymptomatic cancers in China provides a greater than 90 per cent survival at 5 years, emphasizing the importance of early diagnosis.

Once overt symptoms occur, the average survival without treatment is 9 months. The most common cause of death is bronchopneumonia. Patients usually procrastinate in reporting their symptoms, leading to a further delay in diagnosis of often 3 to 4 months. Metastases to lymph nodes and visceral organs as well as mediastinal invasion have often already occurred precluding the chance for a surgical cure. Mass screening programmes in areas of low incidence are not cost effective.

The current TMN staging system was developed by the American Joint Committee on Cancer (AJCC) in conjunction with the International Union Against Cancer (UICC) (Table 3) 295. The primary determinants of survival in surgically treatable oesophageal cancer are the depth of invasion and the presence of lymph node metastases. This staging system incorporates these two critical variables. Survival is stage dependent. Stage I cancer is rare in the Western world. Although postsurgical staging can be accurately performed, presurgical staging is still suboptimal because computed tomographic scanning does not reliably determine the depth of invasion by tumour or the status of the regional nodes. Endoscopic ultrasound may play an important role in the future in this regard.

Dysphagia, first to solids then liquids, is the most common presenting symptom. This is a late symptom and usually indicates that two-thirds of the circumference is involved. Patients often adjust their diet to soft solids and flush their foods through with liquids to accommodate their dysphagia. Weight loss is common and at times disproportionate to the duration of dysphagia. Odynophagia or painful swallowing is often seen. Steady deep chest pain often indicates mediastinal invasion.

Cough or hoarseness is common in upper oesophageal tumours as a result of a direct invasion of the airway or the recurrent laryngeal nerve. Aspiration pneumonia may be present either as a result of overflow obstruction or due to a malignant oesophageal–airway fistula. Patients with a fistula usually have severe paraoxysms of coughing immediately after swallowing liquids.

Physical examination is usually unrewarding. Cervical adenopathy is rarely palpable. Evidence of weight loss is usually apparent.

Laboratory data are rarely helpful in making the diagnosis, although they may be of some benefit in staging (i.e. elevated alkaline phosphatase for bone or liver metastases). Laboratory data also help in assessing the risk of oesophagectomy (i.e. renal and hepatic function). Pulmonary function testing and arterial blood gas measurement are helpful to quantify the extent of chronic obstructive pulmonary disease. Patients with underlying coronary artery disease should have an echocardiogram to assess left ventricular function and a stress test to assess the extent of ischaemic myocardium.

Chest radiographs are usually unremarkable. Possible abnormalities include an oesophageal air–fluid level, infiltrates suggesting aspiration, pleural effusion suggesting pleural dissemination, mediastinal widening (due to the tumour or lymphadenopathy), and pulmonary nodules suggesting lung metastases.

The barium swallow is the mainstay of radiological diagnosis (Fig. 1) 917. Abnormality of the oesophageal axis (angulation of a long axis of the oesophagus) strongly suggests mediastinal invasion and thus helps predict unresectability.

Computed tomography (CT) displays the oesophageal wall and mediastinal structures clearly and in addition, allows evaluation of the common relevant metastatic sites (liver, lymph nodes, adrenal glands, lungs, and kidneys) (Fig. 2) 918. All patients should undergo CT evaluation, but CT does not accurately predict invasion. Gross distortion of a mediastinal structure indicates invasion; the presence of a fat plane between the tumour and the structure predicts the absence of invasion. Evaluation of tumours at the gastro-oesophageal junction is very difficult although it is improved by distension and contrast opacification of the stomach. Evaluation of lymph nodes is suboptimal; enlargement does not always predict metastases and most excised positive lymph nodes are of normal size.

Bone scans are performed for clinical suspicion of bone metastases based on either a history of bone pain or an elevated alkaline phosphatase. Liver scans are inferior to CT in determining the presence of liver metastases.

Oesophagoscopy is required in the assessment of all patients with dysphagia and allows histological (or cytological) confirmation of suspected carcinoma. It is important to measure the length of the lesion and the distance from the incisors for staging and treatment planning. Typical tumours are friable and bleed easily. Multiple biopsies from suspicious areas should be performed. The ability to make a histological diagnosis is increased by taking multiple biopsies. Brush cytology is often very helpful. About 10 per cent of biopsies are non-diagnostic and repeat oesophagoscopy and biopsy is required. The stomach, pylorus, and duodenum should also be examined to assess any additional pathology and to evaluate the stomach for use as an oesophageal substitute.

Bronchoscopic examination of the airway is necessary in all upper and middle third carcinomas. Transmural spread of tumour can be confirmed by visual inspection and biopsy. Bulging of the airway indicates abutment, but does not usually signal direct invasion of the airway.

Laparoscopy can allow reliable assessment and biopsy of the superficial aspect of the liver, perigastric lymph nodes, and the peritoneum, if indicated by previous staging examinations.

Thoracoscopy can be used to confirm invasion into local structures or to sample lymph nodes for staging. The role of video-assisted thoracoscopy for oesophagectomy needs to be defined.

Endoscopic ultrasound improves the ability to determine wall penetration and abnormal lymph nodes. Thus endoscopic ultrasound may improve preoperative staging. Five distinct wall layers can be identified that correspond to the mucosa, lamina propria, muscularis mucosa, muscularis propria, and adventitia. Carcinoma appears as an irregular hypoechoic mass (Fig. 3) 919. Depth of penetration of the wall can be accurately assessed. Regional lymph nodes can be identified and metastases predicted on the basis of size and appearance. The use of endoscopic ultrasound is limited to those neoplasms that allow passage of the probe. However, the presence of a tight malignant stricture is an excellent (90 per cent) predictor of a stage III tumour. Its exact role in the management of oesophageal cancer is uncertain. Its greatest use may be in improved staging for patients undergoing preoperative adjuvant therapy.

The goal of treatment in carcinoma of the oesophagus is twofold: palliation of dysphagia and cure of the cancer. The standard of therapy is oesophageal resection. Resection quickly restores swallowing ability to normal and palliates dysphagia. As a single therapy, surgery offers the greatest chance for cure. Cure rates vary from institution to institution. A cumulative review of all published reports in English for surgical resection of squamous cell carcinoma of the oesophagus by Muller demonstrates the magnitude of the problem and summarizes the experience with 76900 patients. Only 56 per cent of patients have resectable disease at first presentation. Resection was associated with an operative mortality of 13 per cent. Survival was 27 per cent at 1 year, 12 per cent at 2 years, and only 10 per cent at 5 years. Recent reports from selected institutions have shown a decrease in operative mortality and increase in 5-year survival. Many centres have reported mortality rates under 5 per cent and 5-year survivals around 20 per cent with surgery alone for squamous cell carcinoma of the oesophagus.

There are many surgical approaches available for oesophageal resection. The main determinants of the operation chosen are the surgeon's preference and level of the tumour. The three most common approaches currently in use are the Sweet left thoracoabdominal approach, the Lewis laparotomy and right thoracotomy approach, and the Grey–Turner transhiatal oesophagectomy as popularized by Orringer. Excellent clinical results have been obtained for each technique in centres experienced in their use. There is little difference in morbidity or mortality between any of these approaches. Other techniques include the radical en-bloc oesophagectomy, an exclusive left thoracic approach (the stomach mobilization is done through a diaphragmatic incision), and the right thoracoabdominal approach.

The Sweet left thoracoabdominal approach is best employed for gastro-oesophageal junction carcinomas or low oesophageal carcinomas. Tumours 35 cm or more from the incisors are ideally suited to this approach. A double lumen endotracheal tube is used to block and deflate the left lung, enhancing exposure. The patient is placed in the right lateral decubitus position. An oblique left upper quadrant laparotomy is performed to explore the gastro-oesophageal junction area and the liver. If there is no contraindication to resection, a thoracoabdominal incision through the sixth or seventh interspace is performed. The diaphragm is incised circumferentially to avoid injury to the phrenic nerve branches. The stomach is mobilized on the right gastric and gastroepiploic arteries. The omentum is divided, preserving the right gastroepiploic artery. The left gastric artery is double ligated. The gastrohepatic omentum is divided with care taken to identify accessory arteries to the left lobe of liver. The hiatus is dissected. A pyloromyotomy or pyloroplasty is then done according to the surgeon's preference. The oesophagus is then dissected in en-bloc fashion from the pericardium to aorta. The level of anastomosis should be at least 5 cm above gross tumour to ensure adequate margins. If a higher level of intrathoracic anastomosis is desired, dissection can be easily performed below the arch of the aorta and then above it, allowing a supra-aortic anastomosis to be made lateral to the aortic arch. If exposure is limited, a second interspace thoracotomy can be made usually through the fourth interspace, allowing better access for anastomosis. The main advantage of this approach is the excellent exposure of the gastro-oesophageal junction and the ease of mobilizing the stomach, especially in obese patients. It offers the best exposure for the most complete abdominal lymphadenectomy. The limiting factor to this exposure is the position of the heart and aortic arch while doing the anastomosis. This can be partially overcome by choosing a second higher interspace as discussed. The largest negative factor about this approach is the fact that the aortic arch is in the way of the surgeon, thus limiting access to the total thoracic oesophagus and encouraging an anastomosis below the aortic arch.

The Lewis right thoracotomy and laparotomy approach is best for mid- or lower-third lesions. An upper midline laparotomy is performed and the upper abdomen explored. If there are not contraindications to resection, the stomach is mobilized as previously described. It is important to enlarge the hiatus to prevent compression of the stomach when it is brought into the chest. As much of the lower oesophagus is mobilized as can be accomplished from the abdomen since this can be difficult through a high right thoracotomy. A table mounted type of retractor facilitates the dissection of the hiatus and lower oesophagus. The patient is then positioned for a right thoracotomy. The right lung is collapsed and the azygos vein ligated and divided. There is excellent exposure to the entire thoracic oesophagus which is dissected from pericardium to vertebral body. Care should be taken to ligate tissues in between the aorta and oesophagus to avoid injury to the thoracic duct. The stomach is then elevated up into the chest and a high intrathoracic anastomosis is made at the apex of the right chest. Care should be taken to avoid pulling too much of the stomach into the chest. If pulled too tightly a relative obstruction can be created at the hiatus. Excessive stomach will tend to fall into the costophrenic sulcus and impair emptying. The advantage to this approach is excellent exposure of the thoracic oesophagus and ease of anastomosis. It is easy to obtain a wide margin on the tumour because of the excellent exposure of the superior aspect of the oesophagus. The main difficulty encountered with the Lewis approach is lack of exposure of the gastro-oesophageal junction and hiatus especially in obese patients.

Transhiatal oesophagectomy is best used to remove upper-third or lower-third neoplasms. It is relatively difficult to remove stage III midoesophageal tumours with this approach. The operation is done in a supine position with a single lumen endotracheal tube. A laparotomy is performed first and the abdomen is explored. The stomach is prepared as an oesophageal substitute. It is helpful to open the hiatus anteriorly as described by Pinotti. This facilitates exposure of the distal oesophagus almost to the level of the carina. Vessels can be ligated under direct vision. The side of left neck is opened and the oesophagus exposed; care is taken to avoid retractor injury to the recurrent laryngeal nerve in the tracheo-oesophageal groove. The upper third of the oesophagus can be dissected under direct vision. If a transmural tumour is present in this area and if additional exposure is required, a partial upper sternal split can be performed as first described by Scannell. The sternal split facilitates exposure in this area. The area around the carina is often difficult to expose and usually has to be done in a ‘blind’ fashion. The oesophagus is removed and the stomach is brought up through the posterior mediastinal oesophageal bed and a cervical anastomosis performed.

Perceived advantages to transhiatal oesophagectomy are the avoidance of a thoracotomy and a cervical anastomosis. Anastomotic leaks occur more commonly in the cervical area, but can be easily drained and are rarely associated with the devastating complications of thoracic leaks. Although avoidance of a thoracotomy allows greater tolerance of the procedure as compared with transthoracic techniques, any advantages are difficult to prove. The major drawbacks are the limited en-bloc resection, the ‘blind’ area around the carina, and the possibility for an increased incidence of injury to nearby structures (thoracic duct, azygous vein, trachea, and recurrent laryngeal nerves). No randomized studies have been done comparing transhiatal with transthoracic oesophagectomy. Non-randomized studies have failed to demonstrate less morbidity, mortality, or hospital stay with the transhiatal technique (Table 4) 296.

Most surgeons favour a drainage procedure with oesophagectomy. A small number of controlled trials have been performed comparing the results of no pyloric drainage procedure versus using a drainage procedure. No clear advantage has been shown. The incidence of gastric outlet obstruction with no drainage has been under 10 per cent, and the need for reoperation has been under 2 per cent. Dumping is a recognized side-effect of a drainage procedure and is one reason that some surgeons prefer not to perform one. If postoperative gastric outlet obstruction fails to respond to conservative measures, reoperation can be difficult. This is the primary reason that surgeons prefer to perform a pyloromyotomy or pyloroplasty.

In 1942, Churchill and Sweet described a triple-layer technique of oesophagastrostomy and conventional en-bloc resection of the cancer and adjacent lymph nodes. Five years later, Richard Sweet published his initial experience with surgical management of carcinoma of the oesophagus in 141 patients. Operating in an era without sophisticated postoperative monitoring devices, mechanical ventilation, or broad-spectrum antibiotics, his results were remarkable: an operative mortality of 15 per cent, anastomotic leaks in 1.4 per cent of patients, and overall 5-year survival of 11 per cent. This served as a standard for many years. Much of the success is directly attributed to the reliability of the anastomosis and remarkably low anastomotic leak rates. Sweet emphasized the details of technique and warned against factors predisposing to anastomotic leak. The lack of an oesophageal serosal layer and the segmental blood supply of the oesophagus make oesophageal anastomosis more demanding than other intestinal anastomoses. Atraumatic handling of the tissues, preservation of the blood supply of both the oesophagus and stomach, avoidance of the use of crushing clamps, lack of tension on the anastomosis, use of fine interrupted sutures, cutting with a knife or other sharp instrument, and tying sutures gently but firmly to avoid cutting tissues are all important details in the performance of an anastomosis. Few modifications have been made from the technique Churchill and Sweet proposed 45 years ago. Preservation of the blood supply is crucial when mobilizing the stomach and oesophagus. The blood supply of the stomach will be from the right gastric and right gastroepiploic arteries. The oesophagus should not be mobilized beyond a few centimetres above the proposed level of the anastomosis to avoid interference with the segmental blood supply. A circle approximately 2 cm in diameter is scored on a portion of the serosa of the stomach (Fig. 4) 920. The circular defect in the stomach should be 2 cm away from the stapled edge of the stomach to avoid compromise of the blood supply. Individual vessels are identified and ligated with fine silk sutures. This minimizes bleeding while the anastomosis is performed and allows for precise placement of sutures.

Interrupted horizontal mattress sutures of fine suture material (we use 4–0 silk) are used to construct the back row of the anastomosis. Corner stitches are placed first, and the remaining sutures are evenly spaced between them. The sutures on the stomach involve the seromuscular layers and those on the oesophagus, the longitudinal and circular muscle layers. The oesophageal sutures should be deep enough to include both the longitudinal and circular muscles of the oesophagus. The sutures should not be tied too tight, to avoid necrosis or cutting through the muscle.

The oesophagus is opened sharply from one corner stitch to the other. The circular button of stomach is removed. The inner layer is completed with simple sutures including just the mucosa of the oesophagus and the full thickness of the stomach. The knots are on the inside, thereby allowing inversion or turning-in of the mucosa of both the oesophagus and stomach. This is accomplished for the entire circumference of the anastomosis. A nasogastric tube is passed into the stomach under direct vision before a single Connell stitch is placed for closure of the final opening. Healing of the inverted mucosa is an important feature in preventing leakage, and the location of the knots on the luminal side minimizes foreign body reaction within the actual tissues of the anastomosis. The outer row is completed using horizontal mattress sutures as described for the back row of the outer layer.

The omentum mobilized with the stomach is placed over the anastomosis anteriorly to provide an additional layer of coverage. The posterior part of the anastomosis lies between the oesophagus and the more proximal stomach. A few sutures are placed between the stomach and the mediastinal pleura to avoid tension on the anastomosis when the patient is upright, particularly if the stomach is full.

Viability of tissues on each edge of the anastomosis is best maintained if trauma is avoided. The edges are never crushed with clamps and, indeed, are handled with forceps as little as is possible. Once the first stitch is placed and tied, traction on it permits placement of the next without the need for grasping the mucosa with instruments. The sutures are tied by positioning the index finger cephalad to the anastomosis, lifting the stomach to the oesophagus, and avoiding pulling down on the fixed and more fragile oesophagus. Delicacy in suturing is especially important for the outer layer of the anastomosis because the oesophagus lacks a peritoneal surface.

A nasogastric tube passed through the anastomosis for a short time avoids distraction at the suture line by a distended stomach. Gentle, periodic irrigation of the tube ensures its patency. Temporary gastric decompression more than compensates for any potentially deleterious effect of an intraluminal foreign body lying against the suture line for a short period.

We have published our experiences with this technique on a consecutive series of 104 patients. There were three postoperative deaths (2.9 per cent). Two deaths were attributable to pneumonia and respiratory failure (patients aged 59 and 73 years). Both patients smoked at least one pack of cigarettes per day preoperatively. The third death was also due to pneumonia and respiratory failure. This 76-year-old patient had an emergency operation for massive gastrointestinal bleeding. Dilatation was necessary in five patients for anastomotic stricture 3 to 6 weeks postoperatively. One dilatation-to-three dilations were required for successful resolution of dysphagia. Delayed anastomotic stricture was not apparent in this group of patients. All patients had postoperative barium swallows. There were no anastomotic leaks, even of localized type. These results have been reported by others using a similar two-layer technique.

Patients are usually returned to the intensive care unit intubated and are electively ventilated overnight and extubated the following morning. Use of thoracic epidural anaesthesia with a combination of narcotic and local anaesthetic is routinely used and greatly ameliorates postoperative pain. Pulmonary artery catheters are not routinely used unless clinically indicated. Fluid status is carefully monitored and an indwelling bladder catheter is employed for several days. All patients receive perioperative antibiotics. Nasogastric tubes are routinely used and are removed when bowel function returns. Chest physical therapy and nebulized aerosols are routinely administered to improve pulmonary toilet. Jejunostomy tubes are placed in high-risk patients or those who are nutritionally depleted. Enteral feeding is begun usually on the second or third postoperative day. A Gastrografin swallow is obtained to check the anastomosis when bowel function returns. An oral diet is then begun and advanced to a soft solid diet.

The most feared complication following oesophagectomy is anastomotic leak. Leak rates as high as 10 to 15 per cent are commonly reported. Many series, however, do report leak rates under 5 per cent. These are usually associated with a two-layer anastomosis. Cervical anastomotic leaks usually present themselves with fever and a painful, swollen neck incision. The neck incision should be reopened and adequate drainage established. Healing will usually occur with provision of drainage, antibiotics, and adequate nutrition. A stricture may develop, requiring several dilatations to restore normal swallowing. Cervical anastomotic leaks are rarely fatal. Intrathoracic anastomotic leaks are much more serious. Mortality rates of 50 per cent or higher are still reported following free intrathoracic leaks. They demand prompt aggressive treatment. If the leak occurs in the first few days following operation, re-exploration is necessary to exclude the possibility of gastric necrosis. If that is found, the stomach should be debrided, closed, and returned to the abdomen. A cervical oesophagostomy should be performed and wide mediastinal and pleural drainage should be carried out. Reconstruction of the gastrointestinal tract can be done at a later date. Leaks detected after 1 week are best treated by immediate dependent drainage. Drainage should be converted to open drainage with a rib resection at an appropriate time. The critical issue is to establish adequate drainage of all infected material immediately after the diagnosis is made. Failure to achieve adequate drainage may lead to erosion of the airway or mediastinum. Small asymptomatic contained leaks can be seen as well. If they are indeed contained, treatment antibiotics, withholding of oral intake, and adequate nutrition eventually lead to healing. At least 1 week should pass before the patient is restudied. Anastomotic leaks remain the source of greatest morbidity and mortality. They require great judgement and immediate action to avoid a fatal outcome. These facts underscore the need for meticulous attention to detail in performing the oesophagogastric anastomosis.

At the Massachusetts General Hospital in the era prior to multimodality treatment of oesophageal carcinoma, survival following resection in all patients was as follows: 2 years, 31 per cent, 3 years 24 per cent, and 5 years 21 per cent. In a large series of Japanese patients who were grouped according to stage, 5-year survivals were as follows: stage I 60 per cent, stage II 30 per cent, stage III 20 per cent, and stage IV 5 per cent. There is no proof that there is any difference between radical en-bloc oesophagectomy, transthoracic oesophagectomy, or transhiatal oesophagectomy even though they obviously involve removal of different amounts of oesophagus and surrounding tissues as well as different degrees of lymphadenectomy (Table 6) 298.

Radiation therapy remains as one of the possible primary treatment modalities for squamous cell carcinoma of the oesophagus. Numerous large studies have been published facilitating comparison of results with surgical treatment. Unfortunately, 5-year survival statistics in many large series in which patients are treated with a curative intent with high dose radiation therapy show a survival of only 5 to 10 per cent at 5 years. The best results with primary radiation therapy obtained are in cervical carcinomas. Local control and survival are greatest in those patients with early stage I or stage II lesions and are poorest with advanced lesions. Radiation therapy usually does not control the local disease. There is an approximate 70 per cent failure rate after radical radiotherapy at the local site. The majority (75 per cent) of the local failures occur in the first year, thus prohibiting effective palliation for the majority of the patients. However, temporary palliation can be achieved with lower doses of radiation therapy which can be of help in patients who are not expected to live long.

Hancock has recently reviewed the results of preoperative radiation therapy in over 1000 patients. Unfortunately, the average 5-year survival was only 6 per cent overall, which is not a significantly different rate from results of either surgical treatment or radiation therapy alone. In a highly selected group of patients who completed both radical radiotherapy and resection, the 5-year survival rate was approximately 14 per cent, not significantly different from historical surgical 5-year survival rates. Additionally, some groups have reported worse results with preoperative radiation therapy, primarily because of the toxicity associated with radical radiotherapy, leading to greater postoperative complication rates. There have been three well conducted prospective trials involving preoperative radiation therapy (Table 7) 299. In no trial was a significant difference noted in median survival or in 5-year survival compared with surgical treatment alone. Preoperative treatment with radiation therapy is not recommended. One well designed randomized trial was reported of postoperative radiation therapy following oesophagectomy and again no significant difference was noted in long-term survival. Routine use of postoperative radiation therapy is therefore not recommended.

Adjuvant preoperative chemotherapy has been reported, the majority of reports being pilot studies using cisplatin-based regimens. Partial and complete response rates up to 66 per cent have been reported. Other agents used included 5-fluorouracil, etoposide, bleomycin, and vincristine. Responders note a significant improvement in their dysphagia. Treatment toxicity in most series has been mild. Complete clinical response is usually associated with residual microscopic evidence of tumour, underlining the need for oesophagectomy. Complete clinical response has been associated with improved long-term survival in most series. The current programme at the Massachusetts General Hospital for squamous cell carcinoma of the oesophagus involves two cycles of preoperative cisplatin and 5-fluorouracil followed by resection and selective postoperative radiation or chemotherapy. A complete response has occurred in 37 per cent, a partial response in 40 per cent, and no response in 23 per cent. Operative mortality was 4 per cent. Seven per cent of the patients had no tumour in the resected specimen. Postoperative treatment included radiation therapy in 27 per cent, chemotherapy in 31 per cent, both therapies in 16 per cent, and no therapy in 27 per cent. Actuarial 5-year survival for the entire group of patients is 41 per cent. Complete responders had a 5-year survival of 68 per cent whereas partial and no responders had a 5-year survival of 20 per cent. A group of 27 patients followed for a minimum of 5 years now has an absolute 5-year survival of 42 per cent. A national trial is under way to randomize prospectively patients in an attempt to verify this observation.

Steiger from Wayne State University reported the first combined modality series. Their treatment programme included 3000 cGy of radiation in combination with chemotherapy (one-half of the patients received 5-fluorouracil and mitomycin-C and one-half received cisplatin and 5-fluorouracil). Their treatment programme had a significant treatment mortality rate, which approached 30 per cent. No evidence of tumour was found in 31 per cent of patients. Those patients having a complete response appeared to have prolonged survival. Others have reported similar results.

Adenocarcinoma of the oesophagus is reported with increasing frequency from many institutions in the United States. Many authors now report more adenocarcinomas than squamous cell carcinomas in surgical series. A recent epidemiological study by Blot in the United States has shown that incidence rates between 1976 and 1987 were fairly stable for squamous cell carcinoma, but increased more than 100 per cent for adenocarcinoma among men. From 1984 to 1987, adenocarcinomas accounted for 34 per cent of all oesophageal cancers among white men. The corresponding percentages for black men, white women, and black women were 3, 12, and 1 per cent respectively. The rate of increase during the 1970s and 1980s surpassed that of any other cancer. The much higher rates among white compared with black subjects suggest that cigarette smoking and alcohol intake, risk factors that are more commonly present in black subjects, are not major risk factors as they are for squamous cell carcinoma of the oesophagus.

Most of the adenocarcinomas are in association with a columnar-lined oesophagus (Barrett's oesophagus). Other possible sites of origin include the superficial mucosal glands which occur in either end of the oesophagus, the deep submucosal glands which are distributed throughout the oesophagus, and areas of ectopic gastric mucosa. Barrett's oesophagus is thought to be an acquired condition due to severe gastro-oesophageal reflux. The development of adenocarcinoma arising in the columnar-lined mucosa was first described in 1952 and the tendency for this columnar-lined epithelium to develop adenocarcinoma was first reported by Naef in 1975. Barrett's mucosa is found in about 10 per cent of all patients who undergo endoscopic examination for symptoms of gastro-oesophageal reflux. Barrett's mucosa can become dysplastic and when high grade dysplasia develops there is a substantial risk for developing adenocarcinoma. Estimates of risk of malignant change in patients with simple Barrett's oesophagus are about one case per 300 patient years from the time of diagnosis. Compared to a matched population, the risk of developing adenocarcinoma in a patient with Barrett's mucosa is approximately 40 times higher.

Dysplasia is present in the majority of patients who undergo resection for Barrett's adenocarcinoma, lending support to the hypothesis suggesting that dysplasia leads to high-grade dysplasia and hence to invasive carcinoma. Dysplasia is graded using criteria established by the inflammatory bowel disease–dysplasia morphology study group. Low-grade dysplasia consists of mildly dysplastic nuclei confined to the lower half of the epithelium. High-grade dysplasia consists of either severe nuclear dysplasia or dysplastic nuclei extending through the entire thickness of the epithelium. High-grade dysplasia is synonymous with carcinoma- in-situ.

Patients with Barrett's oesophagus should undergo surveillance endoscopy approximately every year to screen for the development of dysplasia or carcinoma. Four-quadrant biopsies should be taken every 1 to 2 cm until normal squamous epithelium is reached. Special attention should be paid to subtle alterations in the appearance of the epithelium, as these areas are more likely to show advanced disease. If low-grade dysplasia is diagnosed, surveillance should be increased to every 3 to 6 months and intensive medical therapy instituted. Dysplasia has regressed with medical therapy in the majority of patients so treated. However, carcinoma has developed during intensive medical therapy for low-grade dysplasia. There is no proof that an antireflux operation can prevent adenocarcinoma from developing.

Once high-grade dysplasia develops, oesophagectomy should be performed. Approximately one-half of patients have invasive carcinoma in the resected specimen (Table 8) 300. The majority of patients so treated are either stage 0 or I. The 5-year survival approaches 100 per cent in stage 0 patients.

Eighty per cent of the neoplasms are in the lower third and 20 per cent are in the middle third of the oesophagus. Patients who present themselves with dysphagia have neoplasms that are usually transmural with regional lymph node metastases. There is extensive submucosal spread of the tumour similar to squamous cell carcinoma. Resection remains the best therapeutic option. The entire extent of Barrett's mucosa should be removed at the time of oesophagectomy as second adenocarcinomas occur in residual Barrett's mucosa. The 5-year survival rate with oesophagectomy approaches 10 per cent. Survival rates approach 90 per cent among those who have early cancer limited to the mucosa recognized by surveillance endoscopy.

Dysphagia is the primary clinical symptom of patients with oesophageal carcinoma that requires palliation. Although palliative resection is recommended by many, it is associated with higher postoperative morbidity and mortality compared with those patients who undergo ‘curative’ oesophagectomies. Oesophageal bypass is considered by some surgeons to provide appreciable palliation of dysphagia in those for whom resection is not possible. Morbidity rates of 50 per cent and mortality rates of 25 to 30 per cent are commonly reported. Median survival following palliative bypass is usually less than 6 months. Most surgeons have abandoned this approach. Use of the Nd-YAG laser may relieve dysphagia, but several treatment sessions are usually necessary to restore an adequate lumen. Repeat treatment sessions are not usually necessary as patients usually die before the tumour regrows. There is a low incidence of perforation with this approach and the ability to swallow is usually quickly restored.

There has been a renewed interest in oesophageal bypass tubes. The older rigid tubes, usually inserted by open techniques, were associated with high mortality rates and frequent complications. The newer silastic tubes are inserted by dilating the oesophagus and pushing the tube in place. The incidence of complications, perforation, and high mortality rates have been low in most series. The distal end is ideally situated above the gastro-oesophageal junction to avoid reflux. These prostheses allow swallowing of liquids and some soft solids. This approach allows quick restitution of swallowing and satisfactory palliation. Average median survival is 3 to 6 months following tube insertion.

About 150 cases of oesophageal small cell carcinoma have been reported with striking morphological and biological features similar to small cell carcinoma of the lung. These tumours contain cells with characteristic cytoplasmic granules (neurosecretory type) which are argyrophilic. Because both the lung and oesophagus are derived from the embryological foregut, argyrophilic neurosecretory cells can be expected in the oesophagus. These cells have indeed been found in small cell carcinoma of the oesophagus. Their clinical course is remarkably similar to that of small cell carcinoma of the lung. The tumours are typically fungating and polypoid with surface ulcerations and are commonly found in the middle and lower third. Widespread metastases are common. Multimodality therapy with chemotherapy and radiation is the treatment of choice with little role for surgery. There are a few long-term survivors.

Melanoblasts have been reported in the oesophageal mucosa and are typically scattered throughout the oesophagus. Oesophageal melanosis is a benign condition seen in approximately 5 per cent of people with apparently normal oesophaguses, but is associated with malignant melanoma of the oesophagus in approximately one-third of all reported cases. Melanoma of the oesophagus is a rare tumour and fewer than 150 cases have been reported. The average age is about 60 and males predominate. Melanomas typically are polypoid and often can grow large. The tumour may be black, brown, or grey; most melanomas are ulcerated. Patients present themselves with dysphagia and when first seen appear to have localized resectable disease. Regional lymph node metastases are quite common. Surgical resection is the treatment of choice although long-term survival is rare. Chemotherapy and radiation therapy have not proved effective.

Leiomyosarcoma is the most common sarcomatous tumour of the oesophagus. Males predominate and the typical age at presentation is 60. Most present themselves in the middle or lower thirds of the oesophagus and have a characteristic pedunculated appearance on radiological examination. The tumour often achieves a large size before obstructive symptoms occur. Myosarcomas are typically well localized and resectable and in general do not invade adjacent mediastinal structures until later in their course. Surgical resection is the treatment of choice and has produced long-term survivors. Although radiation therapy may provide palliation, surgery appears to offer superior cure rates.

Autopsy studies confirm that these are extremely rare tumours. Reports have ranged from 1/1000 postmortem examinations to 2/36000 postmortem examinations. Leiomyomas occurring in the oesophagus represent about 10 per cent of all gastrointestinal leiomyomas. Leiomyomas rarely occur in the cervical region, but are equally distributed between the middle and lower third levels. Less than 5 per cent of the leiomyomas are multiple. Leiomyomas are usually found in men (ratio of 2:1). There is a wide age distribution between 20 and 70 years. The tumours are usually intramural and well circumscribed. Unusual configurations of leiomyomas with a horseshoe pattern are not uncommon. Leiomyomas rarely undergo malignant transformation. The growth rate of these tumours appears quite slow as the duration of reported symptoms often can be quite long. Dysphagia and odynophagia are the most common presenting symptoms. They are frequently found as incidental findings during assessment for other gastrointestinal complaints. Barium oesophogram typically shows a smooth semilunar defect with sharp borders and an intact mucosa. Horseshoe type leiomyomas characteristically produce obstruction. Oesophagoscopy should be performed, but typically only a bulging mass is seen. The overlying mucosa is usually intact, and should not be biopsied. All symptomatic leiomyomas should be removed. Small asymptomatic tumours should be left in place. For mid-third lesions, a right thoracotomy is chosen whereas lower third lesions are best approached through a left thoracotomy. The tumour can be enucleated from the oesophageal wall. Care must be taken to avoid entering the oesophageal mucosa. The incision in the oesophageal musculature is closed with interrupted sutures.

Benign polyps of the oesophagus are rare, but are notable for their sometimes dramatic presentation with regurgitation of the polyp into the mouth which can produce airway obstruction. The polyps are usually solitary and are frequently quite long and cylindrical in shape. Because of constant peristaltic action, elongation frequently occurs which then can permit regurgitation into the mouth. Marked dilatation of the oesophagus can occur because of gradual enlargement of the polypoid mass. Polyps are typically composed of vascular fibroblastic connective tissue, covered by normal mucosa. Dysphagia is the predominant symptom, but occasionally the patient will relate a history of intermittent regurgitation of a mass into the mouth. The barium oesophagogram is often diagnostic, showing a long intraluminal filling defect with a rounded lower border. Oesophagoscopy confirms results of barium studies. These lesions usually occur in older men and usually arise from the cervical oesophagus. Treatment is always surgical, both to relieve symptoms and to rule out malignancy. Occasionally small polyps have a base that can be readily seen and the stalk can be endoscopically divided and the base cauterized. Larger polyps require oesophagotomy on the side opposite to the tumour. This can be done through a cervical approach for high lesions or upper sternotomy or lateral thoracotomy for lower lesions. The oesophagus is closed in layers.

Aikiyama H, et al. Principles of surgical treatment for carcinoma of the esophagus. Analysis of lymph node involvement. Ann Surg, 1981; 194: 438–46.

Blot WJ, et al. Rising incidence of adenocarcinoma of the esophagus and gastric cardia. JAMA, 1991; 265: 1287–9.

Burt M, et al. Malignant esophogorespiratory fistula: management options and survival. Ann Thoracic Surg, 1991; 52: 1222–9.

Cusumano A, et al. Push-through intubation: effective palliation in 409 patients with cancer of the esophagus. Ann Thoracic Surg. 1992; 53: 1006–9.

Daniel TM, et al. Transhiatal esophagectomy: a safe alternative for selected patients. Ann Thoracic Surg. 1992; 54: 686–90.

Delarue NC, Wilkins E Wassel, Wong J, eds. International trends in general thoracic surgery, Vol. 4. Esophageal cancer. St. Louis: C. V. Mosby CO, 1988.

Earlom R, Cunha-Melo JR. Oesophageal squamous cell carcinoma. II. A critical review of radiotherapy. Br J Surg, 1980; 67: 457–61.

Earlom R, Cunha-Melo JR. Oesophageal squamous cell carcinoma. I. A critical review of surgery. Br J Surg, 1980; 67: 381–90.

Gignoux M, et al. The value of pre-operative radiotherapy in esophageal cancer: results of a study of the EORTC. World J Surg, 1987; 11: 426–32.

Hancock SL, Glatstain E. Radiation therapy of esophageal cancer. Semin Oncol, 1984; 11: 144–58.

Hankins JR, et al. Carcinoma of the esophagus. A comparison of the results of transhiatal versus transthoracic resection. Ann Thoracic Surg, 1989; 47: 700–5.

Hilgenberg AD, et al. Pre-operative chemotherapy, surgical resection, and selective post-operative radiation therapy for squamous cell carcinoma of the esophagus. Ann Thoracic Surg, 1988; 45: 357–63.

Launois B, et al. Pre-operative radiotherapy for carcinoma of the esophagus. Surg Gynecol Obstet, 1981; 153: 690–2.

Lewis I. The surgical treatment of carcinoma of the oesophagus. Br J Surg, 1946; 34: 18–31.

Mathisen DJ. Seminars in Thoracic and Cardiovascular Surgery—Esophagus. 1992; 4.

Mathisen DJ, et al. Transthoracic esophagectomy: a safe approach to carcinoma of the esophagus. Ann Thoracic Surg, 1988; 45: 137–43.

Moon BC, Woolfson IK, Mercer CD. Neodymium: yttrium–aluminum–garnet laser vaporization for palliation of obstructing esophageal carcinoma. J Thoracic Cardiovasc Surg, 1989; 98: 11–15.

Muller JM, et al. Surgical therapy of oesophageal carcinoma. Br J Surg, 1990; 77: 845–57.

Pera M, et al. Barett;s esophagus with high-grade dysplasia: an indication for esophagectomy? Ann Thoracic Surg, 1992; 54: 199–204.

Rice TW, et al. Esophageal carcinoma: esophageal ultrasound assessment of pre-operative chemotherapy. Ann Thoracic Surg, 1992; 53: 972–7.

Roth, JA, Ruckdeschel JC, Weisenburger TH, eds. Thoracic oncology. Philadelphia: W. B. Saunders, 1989.

Shahian DM, et al. Transthoracic versus extrathoracic esophagectomy: mortality, morbidity and long term survival. Ann Thoracic Surg, 1986; 41: 237–46.

Skinner DB. En bloc resection for neoplasms of the esophagus and cardia. J Thoracic Cardiovasc Surg, 1983; 85: 59–71.

Streitz JM Jr, et al. Adenocarcinoma in Barrett's esophagus: clinicopathologic study of 65 cases. Ann Surg, 1991; 213: 122–5.

Streitz JM, Williamson WA, Ellis FH. Current concepts concerning the nature and treatment of Barrett's esophagus and its complications. Ann Thoracic Surg, 1992; 54: 586–91.

Steiger Z, et al. Eradication and palliation of squamous cell carcinoma of the esophagus with chemotherapy, radiotherapy and surgical therapy. J Thoracic Cardivasc Surg, 1981; 82: 713–19.

Sweet RH. Surgical management of carcinoma of the midthoracic esophagus. N Engl J Med, 1945; 233: 1–7.

Turner GG. Excision of thoracic oesophagus for carcinoma, with construction of extra-thoracic gullet. Lancet, 1933; ii: 1315–16.