Clinical Risk Factors

Two major risk factors for esophageal squamous cell carcinoma are alcohol use and tobacco smoking, according to epidemiologic studies from various countries worldwide. This relationship is dose-dependent and there is a multiplicative interaction of alcohol intake and tobacco use. In a prospective cohort study from the Netherlands involving more than 120,000 people with 16 years of follow-up, the strongest risk for squamous cell carcinoma was alcohol consumption with a 4.6-fold increased risk, whereas combined exposure with smoking increased the risk more than 8-fold. The risk of squamous cell carcinoma decreased with smoking cessation and alcohol abstinence but only after 1 to 2 decades.¹³

In adenocarcinoma, smoking but not alcohol is associated with increased risk. A prospective study of tobacco, alcohol, and risk of esophageal cancer in the United States found an increased risk of squamous cell carcinoma among current smokers compared with nonsmokers (hazard ratio: 9.27, 95% confidence interval [CI]: 4.04, 21.29) and also an increased risk for adenocarcinoma (hazard ratio: 3.70, 95% CI: 2.20, 6.22). For people who consumed more than three alcoholic drinks a day compared with one drink, there was increased risk of esophageal squamous cell carcinoma but not adenocarcinoma.¹⁴ A pooled analysis from an international consortium composed of 10 population-based case-control studies and 2 cohort studies found strong associations between smoking and adenocarcinoma (odds ratio [OR]: 2.08, 95% CI: 1.83, 2.37).¹⁵ However, alcohol was not found to be a risk factor for adenocarcinoma.¹⁶

Gender, age, and race play a significant role in esophageal cancer although the underlying biological mechanisms are unknown. Men are at significantly higher risk for esophageal cancer overall compared with women. As with many cancers, the risk of esophageal cancer increases with age. Whites are at highest risk for esophageal adenocarcinoma whereas blacks and Asians are more at risk for squamous cell carcinoma.

Nutritional and dietary factors have been found to contribute to the risk of both squamous cell carcinoma and adenocarcinoma. Pickled vegetables, processed meat, and other foods containing nitrates have been linked to an increased risk of squamous cell carcinoma. Drinking very hot liquids or eating hot foods such as stewed meat frequently is postulated to cause thermal injury to the esophagus and has been found to be a risk factor for squamous cell carcinoma. Low consumption of fruits and vegetables are risk factors for both squamous cell carcinoma and adenocarcinoma.¹⁷

Obesity, defined as a body mass index of 30 or greater, is a strong risk factor for esophageal adenocarcinoma, as demonstrated in a meta-analysis of epidemiologic studies (OR: 2.78, 95% CI: 1.85, 4.16).¹⁸ This has clear implications for the increasing obesity rates in the United States and could in part explain the rising incidence of esophageal adenocarcinoma. The exact mechanism for this association of obesity is not understood but might reflect an increased propensity for gastroesophageal reflux. Obesity does not appear to be a risk factor for squamous cell carcinoma.

Chemical irritation and underlying esophageal disease are risk factors for esophageal cancer. Factors that are associated with an increased risk of squamous cell carcinoma include caustic lye ingestion and radiation therapy, and the interval between injury and the development of cancer may be considerable. In addition, a history of underlying esophageal disease such as achalasia, Plummer-Vinson syndrome and tylosis are also linked to squamous cell carcinoma. In adenocarcinoma, gastroesophageal reflux of acid and bile are major risk factors that cause chronic inflammation, which then leads to carcinogenesis.

Adenocarcinoma: Role of GERD and Barrett Esophagus

There is a clear relationship between GERD and the development of Barrett esophagus. Esophageal adenocarcinoma frequently arises in Barrett esophagus, a condition in which the normal stratified squamous mucosa of the esophagus is replaced by a metaplastic columnar-lined epithelium that extends upward from the GEJ, generally as a result of injury from chronic reflux. Various lengths of esophagus may be involved. The condition is an acquired, metaplastic process that develops in response to an esophageal mucosal injury that heals in the setting of the inflammatory stimulus of continued gastroesophageal reflux. Barrett esophagus may progress to dysplasia and then malignancy as the dysplastic epithelial cells accumulate genetic alterations.

Gastroesophageal reflux disease is a well-established risk factor for esophageal adenocarcinoma. A population-based case-control study in Sweden demonstrated an OR of 7.7 (95% CI: 5.3, 11.4) for development of esophageal cancer in patients with chronic reflux disease. With longstanding, severe symptoms, the OR for esophageal adenocarcinoma was 43.5 (95% CI: 18.3, 103.5).¹⁹ There was no association seen between reflux and squamous cell carcinoma. Interestingly, the increased risk of esophageal adenocarcinoma existed whether or not Barrett esophagus could be identified, leading the authors to speculate that the area of Barrett esophagus in these patients was overgrown by tumor. This theory has been supported by postchemotherapy studies showing a high prevalence of Barrett esophagus in adenocarcinoma patients.

The role of screening for Barrett esophagus is controversial. A nationwide population-based study in Denmark found that patients with known Barrett esophagus comprised only 7.6% of all diagnosed esophageal adenocarcinoma patients.²⁰ Similarly, an analysis of U.S. administrative data found that only 8% of adenocarcinoma patients had been recognized to have Barrett esophagus prior to their cancer diagnosis.²¹ Therefore, the value of screening endoscopy in the setting of GERD has been questioned by those who point out that Barrett esophagus is uncommon, progression to malignancy is infrequent, and the effect of screening on overall population mortality from esophageal adenocarcinoma appears quite low. A large nationwide case-control study in Sweden found that among patients with adenocarcinoma, 62% had histologic evidence of Barrett esophagus but 40% did not have a history of gastroesophageal reflux.¹⁹ Thus screening endoscopy of individuals with symptomatic gastroesophageal reflux will still miss the substantial proportion of individuals with esophageal adenocarcinoma who do not report reflux symptoms. The American Gastroenterological Association and American College of Gastroenterology guidelines state that there is currently insufficient evidence to recommend routine screening for Barrett esophagus in patients with gastroesophageal reflux symptoms and that the decision to screen a patient should be individualized.^22,23

The prevalence of Barrett esophagus is estimated to be 1% to 2% of the general population.²⁴ The length of time for progression from Barrett esophagus to dysplasia to adenocarcinoma is unknown. Many advocate lifelong endoscopic surveillance for patients with Barrett mucosa with the goal of treating dysplastic changes and thereby reducing the risk of developing adenocarcinoma. Indeed, tumors that are discovered during surveillance appear to be of an earlier stage and therefore to have a higher chance of cure.²⁵ Still, it is not certain whether surveillance reduces mortality; the overall risk of death from esophageal cancer is relatively low even in this high-risk population. Furthermore, surveillance of Barrett esophagus patients with upper endoscopy is fraught with the problem of sampling error when biopsies are performed during endoscopy and to a high degree of interobserver variability in dysplasia grading.

Several centers have reported results of endoscopic surveillance programs that suggest that progression of Barrett esophagus to adenocarcinoma might be less common than was originally thought, at least in the short term. A nationwide population-based cohort study in Denmark followed up with more than 11,000 patients with Barrett esophagus for a median of 5.2 years and found that the annual risk of adenocarcinoma was 0.12% (95% CI: 0.09, 0.15).²⁰ Meanwhile, a meta-analysis of 57 studies involving more than 11,000 patients with nondysplastic Barrett esophagus found a pooled annual incidence rate of adenocarcinoma of 0.33%.²⁶ Patients with low-grade dysplasia had an incidence rate of adenocarcinoma of 0.5% per year, whereas those with high-grade dysplasia have an incidence rate of adenocarcinoma of 3% to 5% per year.

Although there has been much progress in unraveling the relationship between Barrett mucosa, dysplasia, and esophageal adenocarcinoma, there are still many unanswered questions, such as why Barrett mucosa is seen in such a specific demographic pattern, and the specific triggers that initiate the progression to dysplasia and carcinoma. Further study to identify the individuals who are at highest risk and to understand the molecular progression from Barrett esophagus to dysplasia and invasive malignancy could lead to effective strategies for screening or prevention.

Molecular Progression to Adenocarcinoma

Molecular genetic data support the histologic observation that there is a progression from normal epithelium to Barrett esophagus to dysplasia to adenocarcinoma.27–32 Although a clearly defined sequence of genetic alterations leading to adenocarcinoma has not been defined, an accumulation of abnormalities^27,33–35 has been identified in a wide range of genes that regulate proliferation, apoptosis, invasion, metastasis, angiogenesis, growth, and cell cycle regulation. Tumor suppressor genes have been implicated as early events, as loss of cell cycle checkpoints may be permissive for genetic instability, allowing later transformation in the metaplasia–dysplasia–adenocarcinoma sequence.

In the last decade,³⁶ epigenetic modifications have emerged as heritable and fundamental features of most malignancies, including esophageal adenocarcinoma.^37–42 The best-studied epigenetic modification of the DNA is promoter region hypermethylation, an epigenetic modification that is associated with gene inactivation. A meaningful understanding of the molecular events that result in progression to adenocarcinoma will likely require a greater understanding of this phenomenon as it occurs at least as frequently as point mutations. In oncogenesis, hypermethylation is often associated with inactivation of tumor suppressor genes, of genes that suppress metastasis and angiogenesis, as well as of genes that repair DNA. Methylation of DNA occurs mostly at CpG sites in the genome and is catalyzed by a family of three active DNA methyl-transferases that transfer a methyl group from S-adenosyl-methionine to cytosine to form 5-methylcytosine. Because this reaction can be blocked effectively by a drug, 5-azacytidine, which acts as an irreversible inhibitor of the DNA methyltransferases, the therapeutic potential inherent in reversing DNA hypermethylation is significant.⁴³

One of the earliest events that is thought to occur in the molecular progression of Barrett esophagus to adenocarcinoma is inactivation of one of the alleles of the tumor suppressor gene p16 via DNA hypermethylation, loss of heterozygosity (LOH), or mutations. This event is thought to be triggered by chronic inflammation secondary to acid and bile reflux and has been found to occur at the stage of Barrett esophagus with no dysplasia. The p16 gene is located on the short arm of chromosome 9 and is a cyclin-dependent kinase inhibitor which regulates the cell cycle.29,30,44,45 When p16 is inactivated, this promotes phosphorylation of the retinoblastoma protein, leading to proliferation. This clone of cells may expand and subsequently there may be loss of the second p16 allele as a result of LOH and the formation of several p16 null clones. When Barrett cells begin to acquire the hallmarks of cancer, there is clonal expansion of cells that have a selective growth advantage because of the genetic or epigenetic changes they have acquired.

Studies have also documented the importance of p53 inactivation,46–49 the loss of one allele of the tumor suppressor gene, p53, via mutation. Later, loss of the second allele of p53 via LOH may occur, resulting in the inactivation of p53. The gene p53 is located on the short arm of chromosome 17 and is involved in regulating cell cycle control. When cells sustain DNA damage and cannot be repaired, p53 is responsible for inducing apoptosis and preventing the replication of genetic instability. Abnormalities in p53 usually occur when one allele has been deleted (usually via mutation) and the other allele is functionally inactivated, often because of LOH in a two-hit mechanism. Therefore, inactivation of p53 removes the ability to repair DNA damage and leads to the replication of genetic instability. Missense mutations of the p53 gene cause the protein to have a much longer half-life than normal, resulting in accumulation in the nucleus, where its overexpression can then be detected by immunohistochemical staining. However, p53 staining has been found to be inaccurate in some cases, leading to high false-positive and false-negative rates. Meanwhile, detection of p53 LOH appears to be a more accurate marker of p53 gene abnormalities and has been found to be a predictor of progression to cancer. The loss of p53 is thought to be involved in the progression from Barrett esophagus with no dysplasia to low-grade dysplasia. The inactivation of p53 leads to loss of cell cycle regulation, which may promote genomic instability and aneuploidy, leading to additional changes required for progression to high-grade dysplasia and malignancy.

Genomic instability is detected via DNA content abnormalities such as aneuploidy, which refers to gains or losses in parts of chromosomes. Aneuploidy in fact has been one of the most studied markers for neoplastic progression in Barrett esophagus.⁵⁰ Several prospective studies using flow cytometry within a large cohort of Barrett esophagus patients have demonstrated the presence of aneuploidy during the progression from Barrett esophagus to adenocarcinoma. This promotes the formation of multiple clones, especially as the degree of genetic instability increases, and this clonal diversity ultimately leads to the development of cancer.²²

Several recent studies have found that a combination panel of biomarkers is a better predictor of progression to adenocarcinoma than individual biomarkers alone. Recently, a prospective study of Barrett esophagus patients found that a combination of DNA content abnormalities (tetraploidy and aneuploidy), p16 LOH, and p53 LOH provided the best prediction of risk for adenocarcinoma (RR: 38.7, 95% CI: 10.8, 138.5). Patients with all of these findings had at least a 79% adenocarcinoma risk over 10 years, whereas patients with none of these findings had only a 12% risk of adenocarcinoma over 10 years.⁵¹ In another prospective study of Barrett esophagus patients, the combination of genetic instability and clonal expansion predicted progression to adenocarcinoma. The investigators defined the size of the clone (x) as the length of the Barrett abnormality multiplied by the portion of the cells in the biopsy flow cytometry specimens that carry the lesion. Taking into account the sizes of clones assessed in this way, relative risks for progression to adenocarcinoma, in comparison with those without the clonal abnormalities, were determined. For p53 LOH, the relative risk (RR) was 1.27x for an x cm clone (95% CI: 1.07, 1.50) and for aneuploidy/tetraploidy the RR was 1.31x (95% CI: 1.07, 1.60). A 5-cm clone containing either p53 LOH or aneuploidy/tetraploidy was associated with an RR of 4.16 (95% CI: 2.01, 8.95) for progression to adenocarcinoma.³³

There appears to be a diversity of molecular abnormalities in esophageal cancer caused by actual genetic mutations, epigenetic inactivation, and altered cell regulation. The method of identifying and codifying alterations in these genes into a clinically useful paradigm has not yet proved superior to standard histology for predicting outcome, but more complex analyses using sophisticated molecular techniques such as genomic arrays could be helpful in the future. In the meantime, new molecular abnormalities continue to be identified.⁵²

Nonsurgical Management of Early-Stage (Tis, Ia) Esophageal Cancer

Radiotherapy alone may be an alternative for some cases of early esophageal cancer. Hishikawa reported 68 patients treated with radiotherapy alone, dose 60 to 72 Gy with external beam radiation or 55 to 60 Gy plus a brachytherapy boost.¹²¹ Five-year cause-specific survival and locoregional control rate was 79% and 82%. As locoregional control was only 58% for tumors longer than 5 cm in length, it was recommended that those patients be treated with combined chemoradiation. A more recent series of 54 similarly treated patients also demonstrated similar outcomes of 86% and 79%, respectively, resulting in a recommendation that this radiotherapy option may be appropriate for medically inoperable or elderly patients in this situation.^122,123

Endoscopic techniques may be useful for lesions limited to the mucosa. In contrast to radiotherapy including brachytherapy, the effectiveness of PDT is limited to a depth of 4 to 6 mm and the penetration of laser or thermal energy is 6 mm or less. These techniques may potentially be sufficient for the treatment of Barrett esophagus and selected patients with high-grade dysplasia but are not considered curative therapy for invasive disease. Submucosal invasion increases the risk of lymph node involvement, which cannot be addressed by these therapies. In one series, the rate of local recurrence was 29% in patients treated for mucosal confined adenocarcinoma at a median follow-up of 36 months.¹²⁴ In another report, Tis/T1 lesions had a 44% pCR rate to PDT and T2 lesions had a 28% pCR rate to PDT alone, with control maintained in approximately half of complete responders.¹²⁵ Although these techniques can be effective for very superficial lesions, esophagectomy is still considered the standard of care, with radiotherapy an accepted alternative for patients unable to undergo surgery. Endoscopic procedures for high-grade dysplasia, in situ carcinoma, or T1a (carcinoma limited to mucosa) are discussed above.

Endomucosal resection (EMR) may be the optimal endoscopic minimally invasive approach for early carcinoma. In contrast with photodynamic therapy and radiofrequency ablation, this approach allows pathological confirmation of depth of invasion. This confirmation is critical because the greater the depth of invasion, the higher is the associated risk of nodal involvement, which requires more extensive therapy for adequate curative management. This is appropriate therapy for disease confined to the mucosa, where there is a probability of lymph node involvement that approaches 0%. This approach may also be appropriate for submucosa involvement up to one-third of the thickness: lymph node involvement is observed in 45% (23/51) of cases with deep submucosal invasion versus 10% (3/29) of middle-third and 7.5% (3/40) of inner-third cancers.¹²⁶ Therefore, for lesions with invasion beyond the submucosa, standard management including surgical resection is warranted, although EMR may be appropriate under select circumstances with minimal submucosal invasion.

Esophagectomy for Stage I and IIa Tumors

Esophagectomy is the standard therapy. The outcome is favorable, especially for stage I disease, and is demonstrated in Figure 74-1. This procedure has the advantage of pathological confirmation of the extent of disease, to determine whether combined-modality therapy may be warranted. The varying approaches to esophagectomy are discussed in the section on locally advanced esophageal cancer in the following text. Although surgery alone is considered appropriate for patients with stage IIa disease, combined-modality therapy should be considered for appropriate candidates because the long-term survival is less than 50%, and clinical staging may frequently overlook involved lymph nodes discussed previously (Fig. 74-1).

Transhiatal Resection

The transhiatal esophagectomy (THE) is a frequently used approach, “rediscovered” in 1976 by Dr. Mark Orringer, in which the intrathoracic esophagus is mobilized distally through the esophageal hiatus and proximally through a cervical incision. The increased prevalence of adenocarcinoma of the distal esophagus and GEJ has largely been responsible for the widespread popularity among surgeons of the transhiatal approach. Because of their distal esophageal location, these tumors are invariably near the esophagogastric junction and readily accessible for direct-vision dissection through the hiatus. Moreover, the regional lymph nodes for these distal tumors are in the parahiatal and proximal lesser curvature regions, both accessible via laparotomy. The resected esophagus is reconstructed by using the greater curvature of the stomach, vascular pedicle based on the right gastroepiploic artery or long-segment colon, which is passed up into the neck as vascularized grafts to be anastomosed to the proximal cervical esophagus. Although reports exist on the use of jejunum for long-segment esophageal replacement, small bowel is generally not an ideal option for esophageal replacement because of its mesenteric vascular anatomy, unless specialized techniques with vascular augmentation are used.¹⁰⁷ The esophageal-replacement conduit is passed through the chest into the neck by one of three routes: (1) subcutaneous, (2) substernal, or (3) posterior mediastinum. The posterior mediastinum is the preferred route when possible. The advantages of THE include avoiding postthoracotomy discomfort, wide proximal esophageal margins to ensure complete resection of tumor and Barrett mucosa, cervical anastomosis where the consequences of anastomotic leak are minimized, and an esophageal reconstruction that results in an excellent quality of swallowing. It is well documented that the approach is acceptable for both benign and malignant esophageal disease.^111,120,127 The disadvantages include inability to visualize middle or proximal third tumors, inability to perform extensive intrathoracic regional lymphadenectomy, potential for injury to intrathoracic and cervical structures, and the need for long-segment esophageal replacement. Randomized trials have confirmed similar long-term survival with the use of a transhiatal approach despite a more limited mediastinal lymph node dissection than can be accomplished via thoracotomy.

The transhiatal approach is safe, well tolerated, and associated with infrequent major complications in experienced hands.¹²⁰ In large series, late functional results have been good or excellent in 73% of surgeries and mortality rates are reported as low as none to 3%.^128,129 Recently, Orringer has reported a series of more than 2000 patients who underwent a THE, with a mortality rate of only 1% in those patients who underwent surgery since 1998, an anastomotic leak rate of 9% and 2% pulmonary complications.¹²⁰ In more than 1000 patients who underwent resection for malignancy, the positive margin rate was only 2%.

A multiinstitutional randomized trial¹³⁰ conducted at the University of Amsterdam and Erasmus University included patients with Siewert I and II tumors, using an experimental arm of transhiatal resection with reconstruction using the gastric remnant compared with a right thoracic-abdominal approach. In this study, lymph node dissection varied with the two surgical approaches. For all patients, periesophageal lymph nodes were removed, as was the lymph node–bearing region along the left gastric artery. Clinically uninvolved celiac nodes were not resected. However, in the RTA group, a much more extensive dissection was performed, including lower and middle mediastinal, subcarinal, and right-sided paratracheal lymph nodes (dissected en bloc). The aortopulmonary-window nodes were dissected separately. Through a midline laparotomy, the paracardial, lesser-curvature, left-gastric artery (along with lesser-curvature), celiac trunk, common-hepatic artery, and splenic artery nodes were dissected, and a gastric tube was constructed. The cervical phase of the transthoracic procedure was identical to the transhiatal procedure, but a left-sided approach was used. Resection via thoracotomy potentially allowed wider resection of tumor and periesophageal tissues. Despite this, the pattern of recurrence was similar: locoregional recurrence occurred in 14% and 12% of patients, respectively; distant recurrence in 25% and 18%; and both in 18% and 19% (P = 0.60). Perioperative morbidity was substantially less with transhiatal resection, especially pulmonary complications, but in-hospital mortality was similar.

Intriguingly, a subgroup analysis of long-term outcome, though showing no suggestion of survival difference for the groups at large, raised the following issue: although survival was excellent for patients with no positive nodes (86% vs 89%) and was 0% for more than 8 nodes in both arms, there was a significant benefit to the more extensive nodal dissection in patients with 1 to 8 nodes, 23% versus 64% (P = 0.02).¹¹⁸ If this difference is real and not just a function of unplanned subgroup analysis, the use of neoadjuvant chemoradiotherapy potentially may minimize the impact via radiation treatment of these undissected nodal beds.

Another randomized trial testing transhiatal resection,¹³¹ conducted in Japan, included only patients with Siewert II and III, in contrast to the trial described above. Transhiatal resection was accompanied by a total gastrectomy with D2 lymphadenectomy (including splenectomy) via a laparotomy. Additional dissection of the lymph nodes along the left inferior phrenic vessels and the paraaortic nodes lateral to the aorta and above the left renal vein was done in curable patients. The procedures were undertaken via laparotomy, and the lower mediastinum was accessed transhiatally. With the transhiatal approach, mediastinal resection included the lower esophagus and only the periesophageal lymph nodes. For the other group randomly assigned to a left thoracic-abdominal (LTA) approach, an oblique incision over the left thorax and the abdomen was made. The same procedure as that for TH was done in the abdominal cavity, including the lymphadenectomy as described above. In the thoracic cavity, a thorough mediastinal nodal dissection below the left inferior pulmonary vein was performed. The study was closed when interim analysis suggested lack of benefit to LTA, including the more extensive mediastinal dissection. In both arms, positive paraaortic nodes were identified in approximately 10% of patients, and positive mediastinal nodes in fewer than 10%.

Ivor Lewis Approach

Partial esophagogastrectomy with an abdominal and right thoracotomy approach was originally described by Welsh thoracic surgeon Ivor Lewis¹³² in 1945. This was designed to optimize exposure of the intrathoracic esophagus, which passes through the upper two-thirds of the chest along the right posterior mediastinum. Once the involved intrathoracic esophagus is mobilized, a partial esophagogastrectomy is performed and the esophagus is replaced by stomach, colon, or (less frequently) jejunum, which is passed into the chest along the esophageal bed, and anastomosed to the proximal esophagus, usually at or above the level of the azygos arch. The advantages of the technique are the excellent exposure of the mid to upper intrathoracic esophagus, dissection of esophageal pathology from the surrounding mediastinal structures, and the ability to do a mediastinal lymph node dissection. The disadvantages, however, are related to the use of a thoracotomy and associated morbidity, with limits on the proximal resection margin and the potential for an intrathoracic esophageal anastomotic leak, which is a more difficult management problem than a cervical anastomotic leak. Reported complications include respiratory problems in 11% to 20%, anastomotic leak in 3% to 7%, and wound infection in 5%. Operative mortality ranges from none to 4%.^{107,110,133–135}

Left Thoracoabdominal Approach

The left thoracoabdominal approach uses a single incision extending from the left chest onto the abdomen; it provides excellent exposure of the lower third of the esophagus and left upper quadrant of the abdomen.¹³⁶ This technique is ideal for patients with limited tumors near the GEJ, especially when the extent of gastric invasion is unclear, because it yields superb exposure and maximizes reconstructive options of the lower third of the esophagus. Respiratory complications are the most common postoperative complications with this approach. At least some degree of atelectasis, usually involving the left lower lung, occurs in most patients. Pneumonia is reported to occur in up to 24% of cases. Anastomotic leaks occur in as many as 12% (mean, 3.7%) of cases, leading to a higher mortality rate. Other complications include atrial fibrillation in 10%, wound infection in 1.5% to 5.2%, and (infrequently) empyema and subphrenic abscess. The reported operative mortality is none to 6.2%.^136,137

Multiple Incisions

Multiple-incision surgical approaches combine the incisional strategies of the standard techniques. Of these, the three-incision approach using a cervical incision (right or left), right thoracotomy, and midline laparotomy, as described by McKeown,¹³⁸ is the most common, and is also referred to as the three-incision, three-hole, total esophagectomy or modified McKeown approach. It combines the exposure of the thoracotomy approach for esophageal mobilization or nodal dissection with the advantages of a cervical esophageal anastomosis. Patient outcome results with this technique are similar to those of other approaches, with reported mortality of 3% to 4% and esophageal anastomotic leak rates of 5% or less.¹³⁹

Radical Resections

The majority of patients with esophageal cancer are first seen with locally advanced (stage II and III) disease. In these patients, survival results are poor with surgery alone. Two approaches attempt to improve survival in these patients. One involves the use of combination therapies, such as preoperative chemoradiation followed by surgery (to be discussed later), and the second involves adding an en bloc, wide-field lymphadenectomy to the standard esophagectomy technique. The esophagus has an extensive regional lymphatic drainage. Arbitrarily, the lymphatic drainage has been divided into three zones or fields—cervical, intrathoracic, and abdominal. Standard esophagectomy techniques involve regional, or one-field, lymphadenectomy. Radical approaches advocate two- or three-field lymphadenectomy in conjunction with esophageal resection and replacement. Hagen and colleagues140,141 believe that proximal hemigastrectomy should also be included as part of an en bloc approach, using colon to replace the resected esophagus. Radical esophagectomy is more complex surgery than standard techniques. This is reflected in morbidity rates as high as 58%.¹²² Nonetheless, 30-day mortality rates as low as 1.6% to 4.3% are reported.^{140,142–144} Survival data using radical esophagectomy techniques are conflicting and, therefore, it is unknown whether there is sufficient benefit to outweigh the increased operative morbidity. Hagen and associates^140,141 reported an improved survival in early-stage tumors using en bloc esophagectomy compared with a standard transhiatal technique. Although prospective, their trial was not randomized. Also, earlier-stage patients were selected for the en bloc approach and, therefore, biased the results and conclusions. A more recent update of their results continues to suggest excellent local tumor control and improved survival.¹⁴¹ In another nonrandomized series, Altorki¹⁴³ reported improved survival in patients with stage III disease with radical esophagectomy compared with standard surgical techniques. Nishimaki and colleagues¹⁴² reported an overall 5-year survival rate of 41% with extended radical esophagectomy for thoracic esophageal cancer. In contrast, Bumm and coworkers¹⁴⁴ demonstrated no difference in overall survival between standard transhiatal and radical THE in which two-field lymphadenectomy is added. Despite the wide surgical dissection with radical techniques, Bhansali and colleagues¹⁴⁵ still documented a 21% locoregional cancer recurrence rate. The determination of which cell type and tumor stage, if any, will benefit from radical surgical techniques has yet to be resolved.

Free Jejunal Interposition

Free jejunal interposition permits proximal segmental esophageal resection and replacement, without the need to resect distal esophagus. For technical reasons related to the microvascular anastomosis necessary to support the jejunal interposition, this technique has been limited to replacement of the cervical esophagus for esophageal, laryngeal, or hypopharyngeal cancers or benign strictures (e.g., lye, radiation). Contraindications include factors that would jeopardize the proposed blood supply to the free intestinal segment, such as advanced age, previous carotid surgery and cervical radiation, or conditions that would interfere with the ability to harvest a suitable jejunal segment, such as peritoneal adhesions or inflammatory bowel disease. In resection of the cervical esophageal segment, the branches of the external carotid artery and external jugular veins are preserved as potential host vessels. A segment of jejunum is selected at least 15 to 20 cm distal to the ligament of Treitz. The specific jejunal segment chosen should have a mesenteric arcade supplied by an adequate-size artery and vein. Approximately 20 to 25 cm of jejunum can be resected, although usually 10 to 15 cm is sufficient. The free jejunal segment is then transferred to the neck, where in an isoperistaltic fashion, the proximal esophageal anastomosis is performed, the arterial and venous microvascular anastomosis carried out to the selected host vessels, and then the distal anastomosis completed.

With this technique, the reported graft survival rate is 85% to 95%, and the operative mortality is 5%. For those patients with successful grafting, 90% are reported to have an adequate swallowing quality. If graft failure occurs, a second attempt will be successful in 50% to 75% of cases.146–149

Minimally Invasive Esophagectomy

With the advent of minimally invasive surgical techniques, an interest developed in applying thoracoscopic and laparoscopic techniques to esophagectomy.¹⁵⁰ Certainly, the techniques of gastric and esophageal mobilization have been well established for other complex minimally invasive surgeries. Minimally invasive esophagectomy (MIE) required that these individual techniques be “spliced” together. Approaches used have included laparoscopic transhiatal resection, combined laparoscopic–thoracoscopic procedures, and laparoscopic creation of gastric tube with thoracotomy and other combinations. The need to convert to an open surgery has been uncommon. The number of lymph nodes removed also appears similar to that achieved with open surgery.¹⁵¹ A steep learning curve exists for these surgeries.

Of late, there has been interest in laparoscopic minimally invasive approaches to esophagectomy.155–155 A variety of MIE approaches exist that use laparoscopic, thorascopic, and robotic approaches. Hybrid techniques have also been performed using these video-assisted methods with the addition of the classic approach through the chest, abdomen, and/or neck. There are numerous cases series, generally with short follow-up, and one significant randomized trial. Luketich¹⁵² from the University of Pittsburgh has described results for more than a thousand patients, 76% with adenocarcinoma and 93% with distal and GEJ lesions. In this patient group, 98% had resection with negative margins, and nodal dissection recovered a median of 21 nodes.

There has been a single randomized trial examining the role of MIE.¹⁵⁵ In this randomized trial, 115 eligible participants had resectable esophageal cancer (cT1–3, N0–1, M0) of the intrathoracic esophagus and GEJ. The primary objectives included the short-term outcomes of pulmonary infections, pain, and quality-of-life end points of minimally invasive resection versus standard approaches. In this trial, 93% had neoadjuvant therapy including radiation prior to resection. The end points of infection, pain, and quality-of-life measures were indeed superior in the group undergoing minimally invasive surgery. The R0 resection rates were similar, 84% versus 92% demonstrating success at primary tumor removal after neoadjuvant chemoradiation, as was the number of lymph nodes retrieved and pathological stage. This matches the conclusions drawn from other single-institution series. Extensive data about long-term survival outcome with this approach is not available, but a recent aggregate analysis¹⁵⁶ of the available data suggests that 1-, 3-, and 5-year survival is similar to the outcome expected for more traditional approaches to esophagectomy. Overall, when comparing MIE with open procedures, anastomotic leaks, stricture rates, and short-term survival seem to be comparable; however, MIE may be associated with a lower intraoperative blood loss, shorter length of stay, and less morbidity, but offset by the longer operative times and costs. A recent review by Schumer et al.¹⁵⁷ nicely compared and summarized the available comparative studies, and found no substantial evidence of differences in rates of major complications, mortality, or lymph node harvest.

Survival After Surgery Alone

Cancer-related survival after surgical resection is discussed separately from the description of individual techniques to emphasize the fact that postesophagectomy survival is a function of stage and not of surgical approach. Several points concerning postesophagectomy survival have now become quite clear. The first is that regardless of whether a thoracotomy or nonthoracotomy technique is used, cumulative postoperative survival is the same, at approximately 20% to 25%. This fact has been underscored most graphically by Muller and associates,¹⁵⁸ who reviewed the world literature to compare overall postesophagectomy survival by technique and showed no significant difference. Hulscher¹⁵⁹ performed a metaanalysis of the English-language literature of transthoracic and transhiatal resection of esophageal cancer and found a higher risk of pulmonary morbidity and mortality with the transthoracic procedure, but a similar 5-year survival of approximately 20%. These investigators also compared limited transhiatal resection with THE with extended en bloc lymphadenectomy in 220 patients with adenocarcinoma of the esophagus. No significant difference was found in survival or operative mortality. Large retrospective series suggest a modest improvement in long-term outlook in recent years, probably the result of lower operative mortality.^117,160,161 Gockel¹⁶² found that for the periods 1985 to 1995 and 1995 to 2005, 5-year survival increased from 15% to 25%, and 30-day surgical mortality improved from 8.3% to 3.1%. In addition, better patient selection may play an important role in improved outcome. For example, Steyerberg et al. have proposed a simple scale based on important co-morbidities, age, neoadjuvant therapies, and esophagectomy volume at the treating hospital, which divided patients into groups with predicted 30-day mortality of under 4% to approximately 20%.¹⁶¹ There continues to be focus on reducing morbidity and mortality.^163,164 It is also expected that more accurate preoperative staging with PET scanning and esophageal ultrasonography may improve surgical outcome by removing some patients with existing gross metastatic disease. The long-term results achieved in some significant prospective trials in the modern era are summarized in Table 74-3.

The second fact is that postoperative survival is stage related. Notably, the majority of patients considered for surgery are found to have stage III disease, and survival for these patients, even with surgery, is poor (approximately 10% to 15%). Hofstetter and associates¹⁶⁵ reported results for 1097 consecutive patients undergoing resection and compared outcomes by stage from 1970 through 1985, 1986 through 1996, and 1997 through 2001. Although median survival increased from 7 to 34 months, surgical mortality decreased from 12% to 6%, and the R0 resection rate increased from 78% to 94%, no meaningful difference was found in longer-term survival according to stage. Three-year survival rates through this period were 63%, 52%, and 44% for pathological stage IIA and were 10%, 18%, and 6% for pathological stage III, which did not represent a statistically significant difference over the time periods. Multivariate analysis showed that survival was associated with complete resection and thorough preoperative staging, and that preoperative chemotherapy used in the later years was associated with increased complete resection. For T1N0M0 adenocarcinoma of the esophagus,^166,167 survival at 5 and 10 years may be closer to 77% and 68% after surgical resection.

The third point is that thus far postsurgical survival is little influenced by whether the patient has a squamous cell or adenocarcinoma histology of their esophageal cancer. Holscher and coworkers¹⁶⁸ documented a postresection survival advantage only for patients with stage I adenocarcinoma. Salazar and colleagues¹⁶⁹ reported no difference in cumulative postoperative survival for patients with squamous cell carcinoma and adenocarcinoma. Finally, despite advances in surgical techniques, postesophagectomy survival has remained remarkably stable over time, emphasizing that improvements would likely require effective combined-modality therapies.

Optimizing Surgical Outcome

Evidence-based surgery uses the treatment outcomes of cost, morbidity, mortality, and quality of life to help physicians, health care administrators, and hospitals determine the most appropriate setting and specific management of patients. Data consistently show that increased provider experience improves patient outcome, lowers complication rates, and reduces cost for complex surgeries. In terms of technical difficulty, length of stay, morbidity, and mortality, esophageal surgery is classified as a complex gastrointestinal (GI) operation. Gordon and associates¹⁷⁰ reported reduced hospital mortality, length of stay, and cost for a wide range of complex GI procedures including esophagectomy. The institution of standardized patient-care pathways, a product of the evidence-based surgery approach, reduced hospital cost for esophagectomies while keeping surgical mortality low (1.3%).¹⁷¹ Indeed, the use of clinical protocols for postoperative care has the potential to substantially reduce operative mortality resulting from a variety of surgical approaches, perhaps to 1% or less.^160,172,173

Dimick and colleagues172,173 documented the importance of surgical volume, hospital experience, and intensive care staffing in optimizing outcome after esophagectomy. These studies underscore the fact that even complex surgical procedures, such as esophagectomy, can be both effective therapy and cost-effective.

In an instructive multiinstitutional analysis174,175 of outcome in Medicare patients over 1998 to 1999, a substantial difference in operative mortality was found to exist based on the number of esophagectomies per year at the institution. This varied from 23% for centers doing fewer than 2 cases per year to 8.1% for high-volume centers performing more than 19 per year. The operative mortality rates reported in this series are higher than those reported in clinical trials, and this may relate not only to particular expertise of the study centers in esophageal cancer treatment but also to patient selection for clinical research and the age of the patients. A subsequent analysis suggested that the variation among hospitals was explained, in part, by lower mortality of individual high-volume surgeons with adjusted operative mortality among Medicare patients of 18% for surgeons performing less than two esophagectomies per year and 9% for those performing more than six.¹⁷⁴ Other studies have since confirmed this finding.^176,177

Chemotherapy Followed by Surgery: Is Outcome Improved by Systemic Therapy?

Promising results for survival improvement reported from numerous phase II trials of preoperative chemotherapy regimens tested in newly diagnosed patients178–185 prompted two initial large randomized trials with differing results.^115,186 These data are summarized below. Although preoperative chemotherapy can be justified based on an advantage demonstrated in the Medical Research Council (MRC) trial, there remains some uncertainty. In fact, a US intergroup trial failed to demonstrate either improved local control or survival with similar preoperative chemotherapy compared with surgery alone. By contrast, there are data indicating survival benefit and improved local control when radiotherapy is added to chemotherapy in a concurrent fashion prior to surgery. Hence, trimodality therapy is favored in the United States over the preoperative chemotherapy approach that is preferred in the United Kingdom.

The US intergroup trial115,187 included 467 patients randomly assigned to receive (1) three courses of cisplatin 100 mg/m² plus infusional 5-FU 1000 mg/m² days 1 to 5 before surgery and two courses after surgery (total, five courses) or (2) immediate surgery. No benefit was demonstrated for the addition of chemotherapy in this trial, 45% of patients had squamous cell carcinoma and 55% had adenocarcinoma. There were no differences in resectability or median, 1-, 2-, or 3-year survival rates between treatment groups and between histologic types. The pCR rate was 2.5%. Median and 2-year survivals with and without chemotherapy were 14.9 months and 35% versus 16.1 months and 37%. In both arms, approximately 60% of enrolled patients underwent a gross total resection (R0) and 17% of those patients had a subsequent local recurrence for an ultimate 57% failure to control local disease. There were no differences in surgical morbidity and mortality rates (6%% for both arms). Most patients did not receive the planned two cycles of postoperative chemotherapy: 52% received one cycle and 38% both cycles. An update with longer follow-up reported a 32% disease-free survival rate at 5 years for patients having complete resection and negative margins (R0), whereas only 5% of those with a lesser resection (R1 or R2) were alive.¹¹⁵

The MRC¹⁸⁶ conducted an 802-patient randomized trial, testing two cycles of cisplatin 80 mg/m² and 5-FU 1000 mg/m²/d continuous infusion for 4 days given prior to resection of squamous cell or adenocarcinoma of the esophagus. Patients were required to have resectable tumor, although the staging evaluation was not proscribed by the study. Two-thirds of patients had adenocarcinoma. Median survival was significantly improved from 13.3 to 16.8 months with 2-year survival improved from 34% to 43%. No information about patterns of failure was reported, but similar to the US intergroup trial 57% had a complete resection with negative margins (R0). There is no immediate explanation for the difference between the results of this trial and the US intergroup trial, but these conflicting results may relate to the greater power, less rigorous staging, chance, or unknown differences between the populations in the two studies.

Another trial conducted by the Medical Research Council, the MRC Adjuvant Gastric Cancer Infusional Chemotherapy (MAGIC),¹⁸⁸ demonstrated a survival benefit for three cycles of preoperative epirubicin, cisplatin and 5-fluorouracil (ECF) and three cycles of postoperative ECF compared with surgery alone. This trial enrolled patients with gastric, GEJ, and distal esophageal adenocarcinoma (26% of patients). An update reported a persistent 5-year survival benefit of 23% versus 17% (P = 0.03).¹⁸⁹ Although not powered for subsite analysis, there did not appear to be heterogeneity in treatment effect for distal esophageal lesions or gastric lesions.

Concomitant Chemotherapy and Radiotherapy Followed by Surgery: Addressing Both Local and Distant Failure to Improve Outcome

The rationale for neoadjuvant chemoradiation or trimodality therapy is the high rate of both local and distant failure seen with surgery alone, such that intensified local and systemic therapies are needed to improve survival outcome. For example, data from the surgery alone control arm of the US Intergroup trial (1990-1995)¹⁸⁷ showed a 57% failure to control local disease (including 41% failure to resect all local disease) and 50% distant failure rate in those who did have complete resection.⁴ A single-institution randomized trial conducted at the University of Michigan¹⁹⁰ demonstrated similar survival, local control, and distant first failure rates with surgery alone. Another single-institution study of adenocarcinoma of the distal esophagus and GEJ examined patterns of failure after surgery alone. Thirty-four percent had distant failure, 19% had local failure, 14% had locoregional nodal failure, and 6% had peritoneal seeding.¹⁹¹ These studies show that the rate of complete resection with negative margins and ultimate local control is not improved by the addition of chemotherapy alone prior to surgery. Similarly, local recurrence rate is high (nearly 50%) after definitive chemoradiation.^192,193 However, the trimodality approach of chemoradiation followed by surgery results in a significant improvement in local control compared with surgery alone in trials of patients with squamous cell carcinoma and adenocarcinoma.^190,193

A number of randomized controlled trials of preoperative chemoradiation compared with surgery alone had been conducted with suggestive but sometimes conflicting results, but a recently reported well-powered trial has now provided convincing supporting evidence. Early randomized trials tested chemoradiation with 5-FU/cisplatin-based regimens followed by surgery compared with surgery alone. Most of these trials were underpowered and the results varied. The results of these trials are summarized in Table 74-8. Two trials, those of Nygaard¹⁹⁴ and LePrise,¹⁹⁵ used a sequential rather than concurrent chemotherapy and radiotherapy design followed by surgery and failed to demonstrate any significant difference in median, disease-free, or overall survival. These trials will not be further discussed. Notably, the studies published by Walsh,¹⁹⁶ Urba,¹⁹⁰ and Tepper¹⁹⁷ included patients with adenocarcinoma. Those of Bosset,¹⁹³ Burmeister,¹⁹⁸ Walsh,¹⁹⁶ Urba,¹⁹⁰ and Tepper¹⁹⁹ compared concomitant cisplatin-based chemotherapy and radiotherapy followed by surgery with immediate surgery. Walsh and Tepper demonstrated a significant survival benefit whereas Urba reported a significant improvement in local control but the observed survival benefit was not significant. Bosset and colleagues showed an improvement in disease-free survival and fewer deaths from esophageal cancer but not improvement in overall survival. The trial of Burmeister¹⁹⁸ did not demonstrate a survival benefit but the chemotherapy and radiotherapy are today considered suboptimal. The trials (Table 74-8) reported similar 3- to 5-year survival rates of 32% to 39% for the investigational combined treatment (trimodality) groups whereas the survival of patients in the surgery control arms varied from 6% to 16%. This difference in outcome in the surgery-alone control arms may relate to selection factors and the rigor of baseline staging. The details of these trials, including the recently reported more definitive CROSS⁵³ trial, are described below.

The trial reported by Walsh¹⁹⁶ was limited to adenocarcinoma of the distal esophagus/GEJ and demonstrated significantly improved survival with concurrent 5-FU 15 mg/kg on days 1 to 5, 30 to 35, cisplatin 75 mg/m² on days 7 and 37, and radiotherapy (RT) 40 Gy in 15 fractions on days 1 to 19. The 3-year survival rate was 32% versus 6% and median survival was 16 months versus 11 months (P = 0.01). The complete response rate for all patients enrolled in the preoperative chemoradiation arm was 22%, and there was evidence of nodal downstaging, with 82% node positive in the surgery arm in contrast with 25% after neoadjuvant therapy (P < .001). This study has been criticized for the relatively small sample size (it closed when an early stopping rule was met), for unexpectedly poor results in the surgery alone arm, and for lack of uniform preoperative staging.

The trial reported by Urba¹⁹⁰ randomly assigned 100 patients to surgery (transhiatal esophagectomy) with or without preoperative chemoradiation (cisplatin 20 mg/m² on days 1 to 5, 17 to 21, vinblastine 1 mg/m² on days 1 to 4, 17 to 20, 5-FU 300 mg/m² on days 1 to 21, and RT 1.5 Gy twice a day to 45 Gy). Median survival was 17.6 months with neoadjuvant therapy and 16.9 months with surgery alone, and 3-year survival was 30% versus 16% respectively (P = 0.18). Survival of patients in the trimodality arm was consistent with the Walsh study. The gross total resection rate was in excess of 90% in both arms. The pCR rate was 28%, and did not differ by histology but the number of patients was small (75 adenocarcinoma, 25 squamous cell). Preoperative therapy reduced the incidence of locoregional failure as the site of first failure from 42% in the surgery control arm to 19% (P = 0.0002). Distant failure was not affected, approximately 60% in both arms. Although the improvement in overall survival from 16% to 30% is in an expected range of 10% to 15%, the study was powered to show a much larger difference based on the results of prior phase II trials from this group.

Bosset¹⁹³ reported the results of a multicenter trial limited to squamous cell carcinoma, stages I and II. These investigators found no difference in overall survival with preoperative therapy although there was a significant improvement in disease-free survival and local recurrence-free survival. The regimen was cisplatin 80 mg/m² given 0 to 2 days prior to each set of RT treatments and the radiotherapy consisted of two 1-week courses of 18.5 Gy in five 3.7-Gy fractions beginning on days 1 and 22. Postoperative mortality was significantly higher, 12% versus 4% in the combined treatment arm. This trial has been criticized because chemotherapy was often not administered on the same day as radiotherapy, the chemotherapy was less intensive than in the other trials, and an unusual hypofractionated and split-course regimen of radiotherapy was used.

The Trans-Tasman Radiation Oncology Group (TROG) and the Australasian Gastro-Intestinal Trials Group (AGITG) randomly assigned 256 patients to surgery alone or to one cycle of preoperative cisplatin 80 mg/m² on day 1 and fluorouracil 800 mg/m² on days 1 to 4, with concurrent radiotherapy, 35 Gy given in 15 fractions.²⁰⁰ Sixty-two percent had adenocarcinoma. No survival benefit was identified but treatment was less intensive with only one cycle of chemotherapy given and a lower radiation dose, although there was a suggestion of benefit for patients with squamous cell carcinoma.

An adequately powered US intergroup trial¹⁹⁷ was initiated to definitively assess survival outcome with preoperative cisplatin and 5-FU chemotherapy and concurrent radiation in comparison with surgery alone. Unfortunately, this trial was closed early as a result of poor accrual likely because of the existence of a strong bias among physicians about the benefits of preoperative chemoradiation, lack of consensus about an optimal preoperative regimen, and patient resistance to randomization. Five-year survival was 39% (21%, 57%) versus 16% (5%, 33%), P = 0.005. There was no appreciable difference in surgical complications.

More definitive results in support of combined-modality neoadjuvant therapy are now available from the CROSS⁵³ trial, which accrued patients in 2004 to 2008. Three hundred sixty-eight patients with esophageal and GEJ tumors (82% distal esophagus and GEJ) were randomly assigned to surgery alone or an alternative chemotherapy on days 1, 8, 15, 22, and 29 using carboplatin (AUC 2 mg/mL/min) and paclitaxel 50 mg/m², along with RT 4140 cGy, a regimen of radiochemotherapy 1 week shorter than the other commonly used approaches. No postoperative chemotherapy was given. Seventy-five percent of patients had adenocarcinoma, 24% had GEJ tumors, and an additional 58% had distal third lesions. Patients were well staged using endoscopy, including ultrasonographic and CT scanning. Ninety-nine percent were clinical stage T2 or greater, and 65% were node positive based on clinical evaluation. No adjuvant chemotherapy was planned.

The findings of the CROSS trial suggest successful downstaging with this regimen. One or more positive lymph nodes in the resection specimen were found in 50 patients (31%) in the chemoradiotherapy–surgery group, as compared with 120 patients (75%) in the surgery group (P < 0.001). Fifty-one percent had pathological T2 and greater disease after neoadjuvant therapy, whereas this was the case for 92% of patients treated with surgery alone. An R0 resection was achieved in 148 of 161 patients (92%) in the chemoradiotherapy group, as compared with 111 of 161 (69%) in the surgery group (P < 0.001). A pCR was observed in 23% of patients with adenocarcinoma.

This trial demonstrated that with this regimen of neoadjuvant chemoradiation, outcome was significantly improved to 49 months median and 47% 5-year survival compared with median and 5-year survival of 24 months and 34% with surgical therapy alone. A benefit was observed for both adenocarcinoma and squamous cell carcinoma: for adenocarcinoma, the adjusted hazard rate was 0.741 (0.536, 1.024), P = 0.07. Pattern of failure data are not available, but this study in combination with the previously reported evidence supports neoadjuvant chemoradiation as a standard approach including GEJ adenocarcinoma although definitive evidence specific to this subgroup is still not yet available.

A randomized phase II trial conducted by the Eastern Cooperative Oncology Group,¹¹³ including adenocarcinoma patients only, attempted to assess whether there might be evidence that CPT-11 or paclitaxel-containing platinum-based chemoradiation regimens might lead to superior outcome. Regimens tested included cisplatin 30 mg/m² and irinotecan 50 mg/m² on days 1, 8, 22, and 29 of 45 Gy RT for 5 weeks or cisplatin 30 mg/m² and paclitaxel 50 mg/m² 1-hour infusion days 1, 8, 15, 22, and 29 with RT. Postoperative chemotherapy using the same agents was planned, either cisplatin 30 mg/m² and irinotecan 65 mg/m² or cisplatin 75 mg/m² and paclitaxel 175 mg/m²/d every 21 days × 3. With cisplatin/irinotecan, the pCR rate was 15%, median survival 35 months, and 5-year survival 46%. With cisplatin/paclitaxel, the corresponding results were pCR 17%, median survival 24 months, and 5-year survival 34%. Neither of these regimens appeared superior to the other, nor likely to substantially improve upon the results reported for the regimen used in the CROSS trial or the several studies using 5-FU-containing regimens described above, and there continues to be a need to identify more active systemic agents to improve outcome in this disease with a largely metastatic pattern of failure.

Based on these data, we conclude that the use of preoperative chemoradiation is a rational strategy, which significantly improves local control and may improve survival. Cisplatin and 5-FU and paclitaxel with carboplatin given concurrent with standard fractionation RT, total dose 41.4 to 50.4 Gy is most commonly used. Other platinum-based regimens containing docetaxel and irinotecan have been tested in phase II trials, and pCR rates, preliminary survival outcome, and toxicity do not appear to differ significantly from the other regimens. Therefore, all these regimens may be appropriately selected for individual patients.