Surgery for cancer of the oesophagus

S. Michael Griffin and Richard G. Berrisford

Introduction

Oesophageal cancer is one of the most challenging conditions confronting the surgeon. Resection of the oesophagus requires surgical experience within chest, abdomen and neck. The approach, extent of resection and method of reconstruction require a versatile skillset to which minimally invasive skills may now be added. Oesophageal resection, even in the best of hands, carries the highest risk of morbidity of any operation, and the management of complications challenges the most experienced of surgeons. Not only is the complete resection of all disease often challenging, restoration of gastrointestinal continuity with a gastric or intestinal graft, maintaining an intact anastomosis, is an exacting discipline.

While treatment for cancer of the oesophagus is multidisciplinary, surgery is still the primary mode of therapy. In the UK, 70% of patients now present with adenocarcinoma of the lower oesophagus or gastro-oesophageal junction, which represent a different disease from the previously more common squamous carcinoma. This chapter discusses the surgical management of adenocarcinoma of the lower oesophagus and the cardia (Siewert types 1 and 2), which are frequently staged and treated as oesophageal cancers, but not subcardial tumours (Siewert type 3), which are described elsewhere (Chapter 7).

Unfortunately, the disease often presents late when increasing dysphagia has developed over several months. As a result of poor fitness or unresectable disease only 30–40% of patients are suitable for radical, potentially curative treatment, whilst the majority receive non-surgical therapies with the aim of palliation. Outcome is strongly stage dependent; whilst early tumours have excellent results with surgery alone, the majority with transmural or node-positive tumours benefit from multimodality therapy, combining surgery with neoadjuvant chemotherapy or chemoradiotherapy (Chapter 9). The multidisciplinary team must exercise judgment in the choice of the appropriate combination of therapies for each individual patient. This will depend on patient age, fitness, symptoms, prognosis and evidence base, as well as the overall stage and histopathology.

Surgical pathology

The vast majority of oesophageal neoplasms are epithelial in origin. Some arise from squamous mucosa, but most arise from metaplastic columnar epithelium, resulting in glandular carcinomas affecting the specialised epithelium in the lower oesophagus. Tumour site and histology are two crucial factors requiring assessment: tumours arising from different sites in the oesophagus vary in their behaviour.

Squamous cell carcinoma arising in the cervical and thoracic oesophagus and adenocarcinoma arising in the thoracic oesophagus and cardia differ in their mode of spread and response to therapeutic modalities. It is essential that the anatomical divisions of the oesophagus are described such that the different therapeutic surgical procedures adopted for tumours at each site can be understood (Fig. 5.1).

Figure 5.1 Anatomical regions of the hypopharynx, oesophagus and gastric cardia.

Surgical anatomy

The oesophagus is a midline hollow viscus, starting at the cricopharyngeal sphincter at the level of the sixth cervical vertebra, entering the chest at the level of the suprasternal notch, traversing the posterior mediastinum and entering the abdomen through the oesophageal hiatus in the diaphragm to join the stomach at the cardia. It bears a close relationship to the trachea and pericardium in front and the vertebral column posteriorly. The vagus and its branches are in close proximity over its entire length. There is no serosal covering. The thoracic duct enters the posterior mediastinum through the aortic opening in the diaphragm. It lies on the bodies of the thoracic vertebrae posterolateral to the oesophagus and between the aorta and the azygos vein. The left atrium and the inferior pulmonary veins lie in intimate contact with the left wall of the lower third of the oesophagus.

The TNM classification (Version 7, released in 2009)¹ combines the salient features of the staging process. This classification has divided the oesophagus into discrete anatomical regions (Fig. 5.1) and is described in Chapter 2.

Preoperative surgical preparation

Meticulous preoperative evaluation to accurately stage the tumour and estimate surgical risk is a crucial prerequisite to successful surgical outcome in this disease (see also Chapters 3 and 4).

Nutritional support

Malnutrition is associated with loss of tissue function, leading to many potential complications during the postoperative period, such as wound breakdown, respiratory failure secondary to poor respiratory muscle function, deep vein thrombosis and infective complications.³

The enteral route is preferred, as there is evidence that increased nosocomial infections occur when the gastrointestinal (GI) tract is not used for nutrition in the pre- and postoperative periods.⁴

Perioperative enteral nutrition with the addition of nutrients that can modulate the immune system, termed ‘immunonutrition’, has been proposed to further reduce postoperative complications and improve outcome.⁵

The routine use of parenteral feeding (total parenteral nutrition, TPN) is contraindicated on general and immunological grounds and should be avoided in order to minimise nosocomial infections and associated sepsis.

Preoperative nutritional support

Patients who present with marked dysphagia are at particular risk of malnutrition and occasionally dehydration; they require urgent active management. In this situation, oesophageal dilatation is high risk, and can affect cure rates if perforation occurs. Stent placement makes dissection and operative decision-making more difficult and displacement can occur. Early feeding jejunostomy (laparoscopic or open) may be the most appropriate option if induction therapy is clearly indicated and primary resection cannot be expedited.

Placement of a feeding jejunostomy at the time of surgery is routine in many units. Although mortality related specifically to feeding jejunostomy is less than 1%,⁶ it is not without morbidity, both in-hospital and longer term (from adhesions). Nevertheless, establishing enteral nutrition in patients with complications following oesophagectomy, when they are most catabolic septic and ill, is a major therapeutic problem and avoiding re-operation has significant advantages.

Routine preoperative and postoperative feeding by jejunostomy in every patient has yet to be proven efficacious on current evidence.⁷ Furthermore, a recent randomised controlled study comparing standard perioperative nutrition with immunonutrition failed to demonstrate any clinical advantage.⁷

A variety of other routes of nutrition have been assessed, with a recent systematic review showing no strong direct evidence to support one particular route.⁸ Nasojejunal and nasoduodenal tubes are associated with a significant rate of dislodgement.⁸

Respiratory care

Optimisation of respiratory function is vital in preventing serious pulmonary complications associated with prolonged surgery and thoracotomy (see Chapter 4). Smoking must be stopped as early as possible, ideally 6 weeks prior to surgery, with the aid of nicotine replacement. Preoperative physiotherapy with coughing exercises and effective use of the diaphragm by restoration of muscle strength through ambulation is encouraged. High-risk patients should also be provided with vigorous physiotherapy with or without bronchodilators prior to surgery. Orodental hygiene should be undertaken, removing any source of chronic sepsis that could disseminate infection to the tracheobronchial tree during intubation. Many surgeons advocate routine use of an incentive spirometer; they are inexpensive, and used properly they help to set goals for patients that can be measured, albeit indirectly.

Prophylactic low-molecular-weight heparin together with antithromboembolism stockings must be provided as soon as the patient comes into hospital to reduce the incidence of thromboembolic complications.

Mental preparation/communication

It is now well established that enhanced recovery programmes can result in better outcomes, reduced length of stay, savings in resources, and improved staffing environment. Elements of such programmes include: preoperative assessment, planning and preparation before admission, reducing the physical stress of the operation, using a structured, goal-oriented approach to recovery, and early mobilisation.

Patients and their families should be introduced to the hospital environment, including the intensive care unit if they are routinely sent there during recovery. They should understand what to expect will happen to them through their inpatient journey. A patient diary and information sheet can help them to follow their own progress and reduce uncertainty. Aspects of their pre- and postoperative management should be shared with them, including pain relief, oxygen and intravenous fluid administration, drains, tubes and nutrition. They should have been counselled prior to admission about their treatment options, paying particular attention to results, limitations and expectations of surgery. The counselling process is greatly enhanced by the involvement of a trained clinical nurse specialist in oesophagogastric cancer.

Detailed perioperative preparation and anaesthetic details are highlighted and explained in depth in Chapter 4.

Surgical objectives

Oesophagectomy for cancer should only be undertaken when a potentially curative R0 resection (complete removal of all macroscopic and microscopic cancer) is expected. Unlike colorectal carcinoma, there is no role for resection in the presence of proven distant metastases (e.g. liver), no matter how localised.

Survival is related to the stage of disease. Patients with stage I disease can expect a 5-year survival of greater than 80%,⁹ emphasising the importance of early detection. Resection alone, therefore, must be the chosen method of therapy in fit patients with T1 tumours of the middle and lower thirds of the oesophagus. In stage III disease, surgery alone produces poor results with prolonged survival for only 10–20% of cases.¹⁰ It appears that both neoadjuvant chemotherapy and radiotherapy provide a benefit for these patients.¹¹ Further randomised trials must be completed in order to outline the optimal therapeutic strategy.

Survival following surgical resection for all stages of tumour has improved over the past 20 years, with morbidity and mortality falling. The reasons for this are listed in Box 5.2 and were well described in the COG Guidance Report on Upper GI Cancers.¹² Many studies have confirmed that results parallel experience in managing this condition,¹³ and poor results occur when experience is limited.^14,¹⁵ There is now overwhelming evidence to confirm the influence of surgeon case volume on the outcome of site-specific cancer surgery.^14–16 Centralisation of oesophagogastric resection in specialist units in the UK has provided sufficient caseload to support strong multidisciplinary teams. Other reasons for improved outcome include better patient selection, earlier diagnosis by open-access endoscopy, surveillance of Barrett’s oesophagus, and improved preoperative, operative and postoperative management.

Box 5.2 Reasons for improved results for oesophageal resection

• Increase in specialist units

• Multidisciplinary approach

• Earlier diagnosis

• Better patient selection

• Improved perioperative management

• Enhanced recovery programmes

Principles of oesophagectomy

Resection of primary tumour

Oesophageal cancer spreads longitudinally in the submucosal lymphatics. It is crucial to obtain accurate information to guide the longitudinal resection margin through endoscopy and endoscopic ultrasound. It is still sometimes difficult to ascertain the length required for clear longitudinal margins, particularly in high lesions, making frozen section obligatory for some patients.

Rules on resection margins

The majority of authors favour a subtotal oesophagectomy to optimise longitudinal margins and take into account submucosal spread of both squamous and adenocarcinomas. There has been much debate around what length of macroscopically normal oesophagus to allow for complete resection margins. In squamous carcinoma this pertains especially to the proximal margin, whereas in adenocarcinoma the distal margin (gastric) is usually the greater concern. Skinner ¹⁷ advocated a minimum resection margin of 10 cm from the palpable edge of the tumour. However, this figure does not take into account the nature, location and pattern of occurrence of the primary cancer. Neither does it discriminate between in vivo margins and margins measured by the histopathologist after shrinkage has occurred during fixation.¹⁸

Primary tumours with multicentric lesions require more extensive longitudinal margins. In squamous cancers, three representative patterns of presentation are encountered (Fig. 5.2).¹⁹ Failure to take these into account may explain high R1 rates (40%) when the oesophageal resection margin is limited to only 4 cm; compared to this, R1 was 17% when the margin was 10 cm. A 10-cm resection margin is a goal to attain in both directions if this is possible. In practice, this rule can rarely be achieved. A 10-cm margin either side of a 6-cm tumour would require an overall length of specimen exceeding that of the normal human oesophagus. It is the authors’ opinion that when only a short resection margin can be obtained through the thoracic exposure, a cervical phase with near total oesophagectomy is advisable.

Figure 5.2 (a) A single cancer. (b) Multifocal cancer. (c) Intramural lymphovascular spread. There is a high risk of positive resection margins in (b) and (c). Shaded areas represent submucosal spread.

Adenocarcinoma of the lower oesophagus commonly infiltrates the gastric cardia, fundus and lesser curve. Extensive sleeve resection of the lesser curve and fundus with the formation of a tubular conduit is necessary to minimise positive distal resection margins. Other studies have demonstrated that patients with microscopically positive margins undergoing palliative resection died of other manifestations before clinical evidence of locoregional recurrence.^20,²¹ A tumour-free surgical margin is therefore not the only important factor to be considered in radical surgery. Nevertheless, it should remain the main goal of every operation.

A clear radial resection margin is equally important and is accepted as an independent prognostic factor for oesophageal cancer, with a definition of R1 being less than 1 mm clear margin.^22,²³ The potential benefits of extended lymphadenectomy, discussed later, only pertain if the primary tumour has been completely excised (R0). Radical en-bloc resection techniques, outlined below in operative description, aim to produce a clear radial resection margin. Roder et al.²⁴ showed a statistically significant difference between R0 and R1 (microscopic residual disease) or R2 (macroscopic residual disease) resections for squamous cell carcinoma in a series of 204 resections with 5-year survival rates of 35% and < 10%, respectively. Lerut et al.²⁵ demonstrated a 20% 5-year survival for R0 vs. zero 5-year survival for R1 and R2 resections in advanced stage III and stage IV adenocarcinomas and squamous cell carcinomas.

Resection of lymph nodes

Lymph node involvement is another independent variable for prognosis, in terms of both locoregional recurrence and survival. Patients with higher nodal burdens have worse outcomes after resection. The Worldwide Esophageal Cancer Collaboration has used pooled data from centres that have undertaken radical lymphadenectomy to report in detail the relationship between nodal involvement and survival.²⁶

Nodal tiers

Lymph node tiers draining oesophageal cancer have been described according to the anatomy of the lymphatic drainage of the oesophagus.^2,^27,²⁸ The extent of lymphadenectomy associated with resection of these tiers is demonstrated in Box 5.1 and Fig. 5.3.

Figure 5.3 Extent of resection and fields of lymph node dissection routinely carried out for cancer of the oesophagus.

One-field lymphadenectomy describes removal of diaphragmatic, right and left paracardiac, lesser curvature, left gastric, coeliac, common hepatic and splenic artery nodes.

Two-field lymphadenectomy describes removal of para-aortic nodes (together with the thoracic duct), right and left pulmonary hilar, paraoesophageal nodes, subcarinal and right paratracheal nodes.

Three-field lymphadenectomy describes removal of the first and second fields along with a neck dissection clearing the brachiocephalic, deep lateral and external cervical nodes, as well as right and left recurrent nerve lymphatic chains (deep anterior cervical nodes).

The fields of nodal dissection should not be confused with the histopathological staging of nodal involvement (see Chapter 3, Table 3.3).

As for many other solid-organ tumours, controversy persists as to the value of lymphadenectomy in squamous and adenocarcinoma of the oesophagus.

In the past some authors felt that lymph node metastases were simply markers of systemic disease and their removal conferred no benefit.²³ Others contended that cure could be obtained in some patients with positive nodes by a radical lymphadenectomy with clear resection margins.²⁹ There has been increasing support for radical lymphadenectomy over the last 5 years.³⁰

The rationale for lymphadenectomy

The arguments for formal radical lymphadenectomy are: optimal staging, improved locoregional control and improved cure rates. Formal radical lymphadenectomy goes hand in hand with radical en-bloc resection; for example, resection of the para-aortic nodes requires the plane of dissection to extend to the aortic adventitia. Negative resection margins (especially circumferential margins) go hand in hand with radical lymphadenectomy.

Optimal staging: Radical lymphadenectomy allows more accurate pathological staging.^25,^31–32 TNM7 relies not only on the finding of positive lymph nodes, but on how many are found (N1, 1–2; N2, 3–6; N3, ≥ 7). Without an adequate lymphadenectomy, the patient is deprived of the most accurate prognosis available, and the patient’s unit is deprived of a quality benchmark. Quality control is not possible as the baseline information, accurate pathological stage, is missing. If an inadequate lymphadenectomy is undertaken, survival will be worse than predicted for stage (the phenomenon of stage migration).

Locoregional tumour control: Locoregional tumour control is an important goal in treating oesophageal carcinoma; recurrent locoregional mediastinal disease can be very difficult to palliate. As Lerut et al. emphasise, long-term palliation is a significant benefit for those patients who are not cured by surgery.²⁵ It is impossible to separate potential benefits of radical lymphadenectomy from radical en-bloc resection, as already mentioned.

Radical en-bloc resection with lymphadenectomy has been associated with prolonged tumour-free survival; this is partly a result of consistently clear resection margins and partly a result of complete removal of involved nodes. Nodal and local tumour recurrence is less common after radical en-bloc resection in a number of retrospective series.^25,^31,^33–35 Single-institution comparative studies have also shown better survival.^36,³⁷ Altorki et al. described a local recurrence rate of only 9.7% and a 33% 5-year survival for node-positive oesophageal cancer patients using a three-field lymphadenectomy and en-bloc resection.³⁸

Locoregional recurrence may be further reduced by induction therapy, whether chemotherapy alone or chemoradiotherapy, with fewer positive nodes as well as an increase in R0 resection, consistent with improved disease-free and overall survival in some randomised trials.³⁹

Improved cure rate: It is extremely difficult to demonstrate that radical lymphadenectomy improves cure rate in a conventional randomised controlled trial. Patients allocated to less than radical lymphadenectomy would be understaged, so no baseline would exist and the groups would not be comparable. Furthermore, most patients now have induction therapy that changes their nodal status during treatment. There are therefore very few randomised trials in the literature.40–42

Despite the limitations, there are indications from the Dutch trial,^40,⁴³ comparing radical transthoracic and transhiatal resection, that patients with a limited number of positive nodes (1–8) had a significantly better survival following radical transthoracic oesophagectomy compared with less radical transhiatal resection. This group of patients would be expected to benefit from extended lymphadenectomy if it conferred survival advantage. Node-negative patients did well and those with a higher nodal burden did poorly, irrespective of the radicality of surgery (see Fig. 5.4). Another smaller, retrospective comparison showed improved survival with radical en-bloc resection compared with transhiatal resection in patients with T3N1 disease with fewer than eight nodes involved.⁴⁴

Figure 5.4 Three routes of oesophageal reconstruction: (1) presternal route; (2) retrosternal route; (3) posterior mediastinal route.

The role of radical lymphadenectomy in early-stage disease remains in question, and depends on the precise depth of invasion of the primary tumour.⁴⁰ There is some evidence that even patients with early-stage carcinoma, in whom a significant proportion can have nodal involvement, could benefit from extensive resection with lymphadenectomy.⁴⁵

The most important recent evidence in support of radical lymphadenectomy comes from the Worldwide Esophageal Cancer Collaboration, who reported on their multi-institution international database comprising over 4600 resections for oesophageal cancer in patients who had not had induction therapy.⁴⁶ They showed that prognosis was highly dependent on the number of lymph nodes involved; patients with more than three nodes involved had a 50% likelihood of systemic disease and patients with more than eight nodes involved had almost 100% likelihood of systemic disease.^30,⁴⁷

Even more important was their finding that survival depended not only on how many nodes were involved, but also on how many were removed at resection.⁴⁷ The number of lymph nodes removed was the third strongest predictor of survival after depth of invasion and number of nodes involved. This finding has been corroborated in an analysis of the SEER (Surveillance, Epidemiology and End Results) database, both for oesophageal cancer⁴⁸ and gastric cancer.⁴⁹ It does appear from these data that the number of nodes resected has an effect on survival.

Harvested nodal counts of 23 ⁴⁷ and 30⁴⁸ have been suggested as the optimal number, with caveats given that this is a goal and not an expected target in all patients. The recommendation for adequate staging for TNM7 is a minimum of six nodes. The AJCC (American Joint Committee on Cancer Staging) suggest 18 lymph nodes as the minimum number for accurate staging.⁵⁰ The authors’ view is that this is too low.

The role of the more extensive three-field dissection in oesophageal malignancy is less clear. Five-year survival rates showed no significant difference between two-field and three-field dissection for lower-third squamous tumours;³² the same pertains to patients with adenocarcinoma of the lower oesophagus. Patients with cancer of the upper thoracic oesophagus (third field) may benefit from dissection in the neck.^24,^38,⁵¹

Summary

There is little justification for oesophagectomy to be performed with intent to cure without any attempt to clear the first level of lymph nodes. Patients with either squamous carcinoma or adenocarcinoma of the oesophagus affecting the upper, middle and lower regions have mediastinal lymph node metastases in over 70% of cases.^21,^31,^32,⁵² Over three-quarters of patients presenting with lower-third tumours have positive upper abdominal lymph nodes.⁵³ In order to perform a potentially curative resection for carcinoma in the middle and lower thirds, a dissection of abdominal and mediastinal lymph nodes is therefore essential.

It is the authors’ opinion that a radical en-bloc oesophagectomy with two-field lymph node dissection is the operation of choice for patients with mid and lower third oesophageal cancer and type I tumours of the oesophagogastric junction.

Method of reconstruction of the oesophagus

Route of reconstruction

After resection of the cervical, thoracic or abdominal oesophagus, one of three main paths can be used for reconstruction: presternal, retrosternal and posterior mediastinal (Fig. 5.4).

Presternal route: This route is mentioned for historic completeness. It is approximately 2 cm longer than the retrosternal route, which in turn is approximately 2 cm longer than the posterior mediastinal route. The only indication for using this route is in the situation of multiple previous reconstructions that have compromised the other two routes.

Retrosternal route (anterior mediastinal): The potential space between the sternum and the anterior mediastinum is easily opened up through effective dissection. There is reported to be a lower incidence of cervical anastomotic dehiscence compared with that of the presternal route. Its major disadvantage stems from the unnatural position of the cervical oesophagus in front of the trachea, which can result in an unpleasant sensation on swallowing.

This route is used for reconstruction following emergency treatment of anastomotic dehiscence or the dehiscence of a gastric substitute that has caused posterior mediastinal sepsis. After incomplete resection (R1 and R2) there is some evidence that a retrosternal conduit would be preferable to the posterior mediastinal route.⁵⁴

The retrosternal route is created by blunt finger dissection through the abdominal and cervical incisions and further developed by insertion of a malleable intestinal retractor. The tip of this instrument is passed up to the neck in direct contact with the back of the sternum. Care is taken not to deviate from the midline. The sternohyoid and sternothyroid muscles are divided in the neck and this allows the passage of the oesophageal substitute easily into the left or right side of the neck. If used for a colonic interposition graft, it is sometimes necessary to resect part of the manubrium/sternoclavicular joint to make room for the colonic graft and anastomosis.

Posterior mediastinal route: This route provides the shortest distance between the abdomen and the apex of the thorax and also the neck.

The posterior mediastinal route is the preferred route of reconstruction in the primary surgical excision of oesophageal cancers.^54,⁵⁵

Gastric or colonic substitutes are easily passed through the posterior mediastinum after completion of the oesophageal dissection in the thorax. No attempt is made to close the pleura after this route of reconstruction.

Organ of reconstruction

Reconstruction with stomach: The method of reconstruction should be kept as simple as possible, to minimise complications. The oesophageal replacement is determined by the site of the primary lesion. The stomach is the preferred option as this organ is easy to prepare and involves only one anastomosis.

The patient is positioned supine and exposure obtained using an upper midline incision. There are five broad principles and practices that must be observed in the preparation of the stomach as an oesophageal substitute:

1. The use of isoperistaltic stomach maintaining vascular supply. The right gastroepiploic and the right gastric artery and veins are vital to the viability of the stomach when used as an oesophageal substitute. The greater omentum is opened and the entire course of the right gastroepiploic artery is carefully identified and preserved. The vascular arcade is interrupted at the junction where the right gastroepiploic artery meets the left. The short gastric vessels are divided and ligated (Fig. 5.5).

Figure 5.5 Main arteries of the stomach and points of division of vessels and stomach for oesophageal substitution.

2. Excision of the lesser curvature. Cancers of the lower two-thirds of the oesophagus require clearance of the lesser curve lymph nodes as well as the left gastric, common hepatic and proximal splenic artery lymph nodes. The left gastric artery should be ligated at its origin and resection of the proximal half of the lesser curvature of the stomach, including the cardia, is performed. The right gastric artery contributes to the maintenance of the gastric intramural vascular network and should be preserved if possible. Although the width of gastric conduit appeared not to impact on outcome in one study,⁵⁶ the authors recommend using a gastric tube of 5 cm width or greater to minimise the risk of ischaemia as described by Akiyama et al.⁵⁷ in the 1970s.

3. Preservation of the intramural vascular arcade. Extensive intramural arterial anastomoses between the vascular arcades of the lesser and greater curvatures exist. This has been well demonstrated by el-Eishi et al.⁵⁸ and Thomas et al.⁵⁹ This vascular network must be preserved during resection of the left gastric area of the lesser curvature and the cardia of the stomach. The extent of the resection of the lesser curvature is determined by a line connecting the highest point of the fundus (Fig. 5.6) and the lesser curvature at the junction of the right and left gastric arteries. This allows the removal of all potentially involved lymph nodes, yet preserves the arterial network to the fundus. There is no evidence that the trunk and descending branches of the left gastric artery running along the lesser curve need to be preserved and, from an oncological point of view, it is important that these are excised with the specimen. Care should be taken to ligate the short gastric vessels away from the greater curvature of the stomach to avoid damage to the intramural network and to preserve the extramural vascular network as well. The right gastroepiploic artery provides an adequate blood flow to maintain vascularity in the region of the fundus, which is the area used for anastomosis.