Chapter 20A. Perspective on Malignant Esophageal Disease

It was a pleasure to receive a letter from Dr Michael Zinner inviting me to write a perspective on the chapters of Dr Simon Law's group from Hong Kong on the diagnosis and treatment of esophageal cancer, Dr David Sugarbaker's group from Boston on the techniques used to resect the esophagus and reconstruct the foregut, and Dr James Luketich's group from Pittsburgh on minimally invasive esophagectomy. Of truth, I have rarely read such well-written and thoroughly thought-out reviews. While reading, I identified areas where I was motivated to comment on the discussion from my personal experience. These usually took the form of helpful thoughts or alternatives. Occasionally, they raised a note of caution or took a controversial point of view.

The reason for the dramatic rise in adenocarcinoma of the esophagus continues to remain a mystery even though it represents the largest epidemiologic change ever recorded for a solid cancer. The rise is largely confined to the Caucasian population and is widely attributed to gastroesophageal reflux disease and its complication, Barrett's esophagus. Some have also evoked the rise in obesity as the cause. Based on sound logic, I still support the hypothesis that potent acid suppression therapy played a role in the epidemiologic change. The incidence of adenocarcinoma began to rise in 1975; the same time acid suppression with H2 blockers became widely available. Over time the H2 blockers were largely replaced by proton pump inhibitors that had a greater capacity to suppress acid. Although this hypothesis has been difficult to prove short of a large and long prospective randomized study, an unbiased review of the small nonrandomized studies published in the surgical and medical literature to date will convince the practitioner of the truth of this hypothesis. A recent study1 showed that medically treated patients who had relief or mildly persistent reflux symptoms while on proton pump inhibitors have significantly higher odds of developing esophageal adenocarcinoma than medically treated patients who have persistent severe reflux symptoms. My explanation for this observation is that acid suppression therapy decreases reflux symptoms by decreasing the acid content of gastric juice and causing the pH to rise from less than 2 to 4 or greater, that is, in the range of a weak acid. On the stealth side, this increase in pH also increases the solubility of bile acids in the refluxed neutralized gastric juice. Bile acids at the pH of 4–6 have ready access into the Barrett's epithelial cell. When in the cell, bile acids are known stimulants of CDX2, the most powerful genetic stimulus for the development of intestinal metaplasia. They also stimulate the expression of genes involved in the carcinogenesis of intestinalized Barrett's epithelium to adenocarcinoma. In contrast, the persistence of severe reflux symptoms while on medication indicates incomplete acid suppression. The persistent acidic gastric juice causes bile acids to precipitate out of solution. This nullifies their effect, hence less or no intestinal metaplasia or adenocarcinoma.

I take exception, as does Dr Law, with the current dogma that there is no evidence proving that surveillance will result in a better survival of patients with Barrett's esophagus who progress to adenocarcinoma. Rather, I believe that all individuals who are identified as having Barrett's esophagus should enter a surveillance program. In our experience surveillance of patients with Barrett's esophagus does detect tumors at an earlier stage. Indeed, the percentage of patients presenting with an early T1 N0 adenocarcinoma has increased over time and, in the more recent years, account for nearly 50% of all resected tumors. Looked at from a different perspective, 86% of esophageal adenocarcinomas identified by Barrett's surveillance have stage I disease.

In the new staging system cancers of the gastric cardia that extend into the gastroesophageal junction are classified as adenocarcinomas of the esophagus rather than the stomach. This is an improvement but still is imprecise due to the inability to consistently define the location of the gastroesophageal junction. In 2000, the Association of Directors of Anatomic and Surgical Pathology defined the gastroesophageal junction as a horizontal line drawn across the end of the tubular esophagus at the point where it begins to flare into the stomach. Using this definition, there is evidence that adenocarcinomas of the gastric cardia commonly arise in areas of intestinal metaplasia within the gastric cardia that have strong similarities to adenocarcinomas of the distal esophagus. We have suggested that the location of the gastroesophageal junction is defined more accurately by histology as the proximal limit of gastric oxyntic mucosa. The area cephalad to this limit is histologically the esophagus and in normal people is lined by squamous epithelium and in patients with reflux is lined, to a variable extent, with different types of metaplastic columnar epithelium (cardiac, oxyntocardiac, and intestinal). Important to understanding disease at this location is to obtain an accurate estimate of the incidence of Barrett's adenocarcinoma. With greater precision it is likely that the number of Barrett's adenocarcinomas would be more than double. This would give us a greater appreciation for the exponential rise of adenocarcinoma in this area and could well make the screening of patients with Barrett's more cost-effective.

Endoscopic surveillance of patients with Barrett's esophagus has identified a rather large number with high-grade dysplasia in the Barrett's segment. Most regard this finding as a threshold for intervention. The new technique of endoscopic mucosal ablation has allowed the treatment of high-grade dysplasia with preservation of the esophagus and a morbidity and mortality lower than an esophagectomy. The survival following either treatment is similar. This has reduced the use of surgical resection to treat these patients. Many, however, have visible lesions within the flat Barrett's segment such as a nodule or ulcer. Such lesions must be removed by endoscopic mucosal resection to determine the nature of the lesion and, if a cancer, its depth of penetration into the esophageal wall. If a cancerous lesion is limited to the lamina propria, ablation of the Barrett's segment can proceed. If a cancerous lesion extends beyond the muscularis mucosa, there is a significant increase in the probability of lymph node metastasis and an esophagectomy is required. Despite flawed statements to the contrary, there is no “safe” level of invasion into the submucosa that would extend the use of endoscopic resection. To manage these patients correctly requires that surgeons become adept at endoscopy and endoscopic mucosal resection. The opportunity for this training is limited and is an issue that must get the attention of the Residency Review Committee for Surgery and the American Board of Surgery. The new therapy is extremely work intensive and is associated with the risk of developing cancer during the treatment. Consequently a vigilant support staff is necessary to handle the patients. In our experience patients who have high-grade dysplasia in a long segment of Barrett's esophagus, or in an anatomically short esophagus with a large hiatal hernia, or in an esophagus with a severe motility problem, or have multifocal high-grade dysplasia or multiple failures of ablation therapy are not candidates for esophageal preservation therapy and are better off having a vagal sparing esophagectomy. This form of esophagectomy is associated with less perioperative morbidity and a shorter hospital stay than a standard transthoracic or transhiatal esophagectomy. Further, its late morbidity, including weight loss, dumping, and diarrhea, is significantly less.2

Surgical resection remains the mainstay of treatment for fit patients with localized esophageal carcinoma that has invaded into the submucosa or beyond. In my mind, the only exception to this rule is a cervical esophageal cancer that is located sufficiently close to the cricopharyngeal muscle to prevent a clear resection margin. These patients are better off receiving definitive radiochemotherapy. If a recurrence occurs, a pharyngolaryngoesophagectomy is performed as a salvage procedure. It is critical that these patients understand this approach prior to treatment and are willing to submit to yearly surveillance after the definitive radiochemotherapy. Fit patients with tumors in the lower cervical or upper thoracic esophagus are treated with cytoreduction of the tumor by radiochemotherapy followed by resection and reconstruction with a free jejuno-interposition. Fit patients with tumors in the mid or lower thoracic esophagus, gastroesophageal junction, or gastric cardia are treated with an en bloc esophagectomy and complete lymphadenectomy. The superiority of this approach has become gradually apparent over the years with the greatest benefit seen in patients with fewer than eight lymph nodes involved in specimens that contain more than 30 resected nodes. The historical development of this position is nicely documented in the 10 publications.3–13

Critical in performing an en bloc esophagectomy is that the proximal, distal, and radial margins are free of tumor. I agree with Dr Law's comment that the proximal esophageal margin is most critical and a good guide is to obtain 10 cm of grossly normal esophagus above the superior margin of the tumor. After removal, the fresh specimen contracts to approximately 50% of its length or down to 5 cm of grossly normal appearing esophagus. With this length of margin, there is less than a 5% chance of an anastomotic recurrence. Similarly, it is important to have a greater than l-mm free circumferential radial margin on the specimen after a curative resection. This has been shown to be an important independent prognostic variable.13 Simply put, patients with less than a 1-cm free circumferential radial margin in what would otherwise have been a curative resection doubles their risk of dying from cancer. As would be expected, the effect of a clear circumferential radial margin is most important in patients who have limited or no lymph node involvement.

An en bloc resection includes a complete lymphadenectomy. The number of lymph nodes removed is an independent predictor of survival. To maximize this survival benefit, a minimum of 23–29 nodes need to be removed. Taking additional nodes is of benefit, but the effect begins to drop off. When analyzed by Cox regression, the number of lymph nodes removed modeled as a continuous variable was the third most important prognostic factor behind the number of involved nodes and the depth of tumor invasion. Of the three factors, the number of nodes removed is the only predictor of survival that can be influenced by the surgeon. The operation most likely to maximize the number of nodes removed is the two-field en bloc esophagectomy.

I am not convinced that adding a third field, that is a cervical node dissection, improves the survival sufficiently to overcome the increased morbidity. Rather, we have taken the approach of obtaining a positron emission tomography (PET) scan and ultrasound examination 1 year after the initial resection and performing a modified radical neck dissection if involved neck nodes are detected or suspected. An exception to this policy is when unsuspected involved recurrent laryngeal nodes are discovered while performing the neck dissection during the initial operation in preparation for a neck anastomosis. In this situation a cervical node dissection is added to the initial operation to remove recurrent laryngeal and deep cervical nodes on the left.

In a unique study, the en bloc and transhiatal resections were compared using a retrospective case-control study of nonrandomized patients with similar-size transmural tumors (T3) and lymph node metastasis selected at random from our registry. The result showed that the survival benefit of an en bloc resection was limited to patients with eight or fewer involved nodes. There was no difference in outcome when nine or more lymph nodes were involved. When this information was applied to the 5-year outcome of the only randomized studies done to compare the two resections, only those patients with one to eight involved lymph nodes significantly benefited from an en bloc resection. This finding fits with the results of a multi-institutional international study showing that the probability of systemic disease is 50% when three nodes are involved and approaches 100% when more than eight nodes are involved. Based on these studies, the en bloc resection is most likely to benefit patients with eight or fewer lymph nodes involved. Beyond this number the likelihood of systemic disease approaches 100%, and neither an en bloc nor a transhiatal resection provides a long-term benefit.

The most dreaded complication of esophagectomy is ischemic injury to the conduit used for reconstruction. This is often the cause for an anastomotic leak and a cascade of sepsis, multiorgan failure, and death. Factors known to contribute to this complication are diabetes, hypertension, cardiac arrhythmia, chronic obstructive pulmonary disease, and neoadjuvant therapy. Indeed, I believe that conduit ischemia is the “Achilles heel” of a successful esophageal resection and reconstruction. When faced with a worrisome ischemic conduit, we pull the conduit up and anchor it in the neck without performing the anastomosis. A Prolene stitch is placed into the conduit and brought out to the subcutaneous tissue for a guide to find the conduit at the time of delayed reconstruction. The proximal esophageal remnant is brought out to the neck as an esophagostomy and a feeding jejunostomy constructed in the abdomen. Ninety days after the esophagectomy, a cervical esophagogastrostomy is performed through the original neck incision. Over the years we have used this strategy in 35 patients. At the time of reconstruction, all had well-perfused gastric conduits and the delayed anastomosis healed without a leak, wound infection, or sepsis.14

As Dr Law pointed out, the past two decades have witnessed a proliferation in chemotherapy and chemoradiotherapy trials in esophageal cancer. The basis for this explosion is the suboptimal surgical cure rate for advanced cancer. True, distant failure remains a major problem in patients with advanced cancer, and a search for more effective systemic drugs as well as a method to select the right drugs for the right patient needs to be supported and encouraged. I agree with Dr Law that currently the results of neoadjuvant chemoradiation therapy are conflicting, and that published meta-analysis show minimal to no benefit. Despite this, neoadjuvant chemoradiation is widely practiced in the United States. A limitation of the current randomized trials is the lack of accurate staging prior to randomization. If randomization is done correctly, major known factors that affect survival, such as stage of disease, need to be evenly distributed in the study groups prior to randomization as the concept of randomization is used only to manage unknown factors that affect survival. I have concluded that the studies done to date have shown neoadjuvant therapy to be only effective in causing cytoreduction of the primary tumor. The lack of a similar response in the secondary lesions is an indication that their sensitivity to chemotherapy is different than the primary tumor, perhaps through tumor cell interaction with host tissues cells and their cellular immune response. I would recommend that future neoadjuvant studies be done only on patients of similar stage based on carefully done pretreatment minimally invasive surgical staging. I would also suggest an adjuvant chemotherapy trial where the chemotherapy given is based on chemosensitivity studies done on the involved lymph nodes removed at the time of surgery or on biopsies of solid-organ metastasis done at the time they are discovered after surgery. Clearly, a new approach is needed. The movement toward definitive chemotherapy is based on its known cytoreduction effect on the primary tumor. The concept is designed to eliminate surgical therapy because surgery as currently performed has failed to control local disease. This is largely due to the resistance and inability of surgeons to perform an en bloc resection. After an appropriately done en bloc resection the local recurrence rate is less than 2% whereas after transhiatal resection it is 25% or greater. My concern is that our failure to centralize esophageal surgery in the United States, as is currently being done in England, will relegate surgical therapy from a primary position in esophageal cancer to an adjuvant role.

Dr Wee and Dr Sugarbaker provide an excellent description of the en bloc esophagogastrectomy done through what is known as the tri-incision or McKeown technique. We perform the operation similar to their description with a few exceptions. We begin in the right chest by dividing the intercostal veins as they join the azygos vein from the arch down to the diaphragmatic hiatus. We then dissect out the intercostal arteries and follow them to where they join the aorta. The aorta is then easily skeletonized from the right and over into the left chest as the left intercostal arteries pass directly posterior to the costospinous junctions and have never interfered with this dissection. The mobilized azygos vein is divided at its junction with the superior vena cava. Both the azygos vein and the thoracic duct are taken with the specimen. The distal azygos vein and thoracic duct are ligated adjacent to the spine deep within the esophageal diaphragmatic hiatus, using a laparoscopic endo-loop. The proximal thoracic duct is divided later in the dissection and does not leak because of proximal valves. Both the right and left sides of the chest are drained using ½-in Jackson-Pratt (J-P) drains positioned adjacent to the spine on the right and aorta on the left and both resting on the posterior chest wall. The drains are placed through the esophageal diaphragmatic hiatus during the abdominal portion of the procedure after the specimen has been removed and before the gastric conduit has been pulled up. The drains are brought out through stab wounds in the right and left upper quadrant. This allows the right and only chest tube to be removed on the first or second postoperative day. We also skeletonize the superior and anterior wall of the common and right hepatic artery, the superior and inferior wall of the portal vein, and the superior and anterior wall of the splenic artery out to the splenic hilum. Skeletonizing the inferior wall of the portal vein is done by using a vein retractor to displace the vein caudally and using the cautery along the superior border of the pancreatic head. The width of our gastric conduit is 3–4 cm, and a pyloroplasty is performed using an end-to-end stapler (EEA, US Surgical, Norwalk, CT) inserted through the conduit staple line near the antrum and taking a cookie bite out of the anterior portion of the pyloric ring.

Dr Luketich's group deserves the credit for being on the frontier of adapting esophagectomy into a laparoscopic and thoracoscopic procedure. Their work is commendable and has shown that there is some reduction in procedural morbidity, postoperative discomfort, and length of hospital stay, but not as much as one would suspect. To their credit they appear not to have limited the extent of the resection to accommodate the new approach but rather creatively altered their approach to maintain the extent of the dissection. There is a point when the benefits of minimally invasive surgery are overcome by the extensiveness of the internal surgical dissection and manipulation. When that point is reached, the advantages of a minimally invasive procedure will diminish and the world of surgery will continue to do such a procedure openly until further technological developments occur that will allow us to go further in our quest for “user-friendly” surgery. I believe minimally invasive esophagectomy is near that point.