36 Cancer of the Esophagus
Epidemiology, Etiology, Genetics, and Cytogenetic Abnormalities
The epidemiology of esophageal cancer has changed during the past decade. There is an increasing incidence of adenocarcinoma, most commonly occurring at the gastroesophageal (GE) junction.1 In 2008, there were an estimated 16,479 new cases of esophageal cancer in the United States, with 12,970 occurring in men and 3500 occurring in women. Approximately 14,280 deaths are expected.2
Squamous Cell Cancer
In African-American men, the incidence is higher than the incidence in White men (16.8 versus 1.0 per 100,000).3 In most countries, the primary contributing factor is tobacco and alcohol abuse.4 Diet can have a protective benefit.5 For example, high intake of raw vegetables and fresh fruit (especially citrus) is associated with a decreased risk.
Adenocarcinoma
In most large referral centers, the incidence of adenocarcinoma of the gastroesophageal (GE) junction represents 60% to 80% of newly diagnosed patients, compared with 10% to 15% only 10 years ago.6 Data from the Surveillance, Epidemiology and End Results (SEER) Program database reveal that the incidence of adenocarcinoma in White men doubled between the early 1970s and the late 1980s. By contrast with squamous cell cancer, the male to female ratio is 7 : 1 and there is an increased incidence in Whites versus African Americans. In a Swedish population-based case–control study, Lagergren et al. reported a strong association between symptomatic gastroesophageal reflux and adenocarcinoma.7
Barrett’s esophagus is a metaplastic change of the lining of the esophagus whereby normal squamous epithelium is replaced by columnar epithelium.8 The pathophysiology, biology, and possible molecular events associated with Barrett’s has been extensively reviewed.9,10 There is strong evidence to support the theory that most adenocarcinomas develop from areas of Barrett’s esophagus.11 Antireflux medications, which relax the lower esophageal sphincter, may be a contributing factor. For example, Lagergren et al. reported that patients who used these agents for more than 5 years had an increased risk compared to those who never used them.12 Interestingly, only 10% of patients who have Barrett’s develop adenocarcinoma; however, most patients with adenocarcinoma have a history of Barrett’s. In fact, most patients with Barrett’s will die of other causes.
Although not as prevalent as reported in patients with squamous cell cancer, tobacco and alcohol abuse also increases the risk of adenocarcinoma.11 A multicenter case-controlled study identified excess weight (body mass index) as a risk factor.13 Likewise, a diet high in calories and fat is associated with an increased risk.14 Although Helicobacter pylori is a risk factor for gastric cancer, it does not appear to be associated with adenocarcinoma of the GE junction.15
Emerging data suggest that the development of esophagus cancer, as with many other epithelial-derived carcinomas, is a multistep process that involves a successive activation or deletion of genes and their protein products. This is mediated by regulatory genes.16–19 Markers such as cyclin D1 are associated with more aggressive disease.20 Activated transcription factor nuclear factor kB is associated with chemoresistance and poor prognosis.21 A comprehensive discussion of molecular genetics is beyond the scope of this chapter.
Anatomy
The esophagus is generally divided into three segments: cervical, thoracic, and abdominal (GE junction). The esophagus can also be classified as the upper third (proximal), middle third, and lower third (distal). Not only is there is no consensus as to the delineation between these segments, but there is overlap between the various definitions.22 Some definitions use the distance from the incisors (Fig. 36-1), whereas others use anatomic criteria such as the thoracic inlet and carina.
Pathology
As seen in Table 36-1, a variety of histologic types of tumors arise in the esophagus.23 The most common are squamous cell and adenocarcinoma. The location of the tumor also influences the histologic type. In the upper and middle third, the majority of cancers are squamous cell, whereas in the lower third they are usually adenocarcinomas. The treatment results discussed in this chapter will be limited to these two histologies. Given the increasing incidence of adenocarcinoma of the esophagus compared with squamous cell carcinoma, treatment results need to be examined by histology. At the time of writing, the data are conflicting, with some series reporting different results by histology whereas other series report no difference. Fortunately, the National Cancer Institute (NCI) Intergroup randomized trials are now stratified by histology. Until these stratified results are available, the impact of histology cannot be adequately assessed and it is reasonable to treat both histologies in a similar fashion.
NOS, Not otherwise specified.
Routes of Spread
The incidence of lymph node involvement increases with T-stage. The incidence is also related to the site of the primary tumor in the esophagus. Metastases from the cervical esophagus to the abdominal lymph nodes are unusual, whereas from the upper thoracic esophagus, metastases may occur in greater than 30% of patients. The relative incidence of metastasis is summarized in Fig. 36-2.24 Hematogenous metastases may occur in 25% to 30% of patients at the time of presentation and in as high as 50% of patients at the time of autopsy. The most common sites of metastasis are lung, liver, pleura, bone, kidney, and adrenal gland. Regardless of the anatomic location or histologic subtype, the median survival of patients with distant metastatic disease is 6 to 12 months.
FIGURE 36-2 • Location of the lymph node involvement (%) by esophageal carcinoma is determined by site of primary tumor.
(From Sons HU, Borchard F: Cancer of the distal esophagus and cardia, Ann Surg 203:188, 1986; Akiyama H, Tsurumaro M, Kawamura T, et al: Principles of surgical treatment for carcinoma of the esophagus, Ann Surg 194:438, 1981.)
Diagnostic and Staging Studies
Diagnosis
The initial diagnosis is often made on upper endoscopy, which allows for delineation of the proximal and distal extent of the tumor and biopsy. The barium esophagram can also demarcate the proximal and distal margins as well as identify a tracheoesophageal fistula. It is helpful for correlation with simulation films. Computed tomography (CT) scan of the chest and abdomen is essential for staging, because it can identify extension beyond the esophageal wall, enlarged lymph nodes, and visceral metastases. Endoscopic ultrasound is highly accurate in determining depth of invasion as well as lymph node involvement.25 Patients with tumors arising or extending above the level of the carina should also be referred for a bronchoscopy to evaluate for a tracheoesophageal fistula. For cervical primaries, a head and neck examination should include palpation of the Level I to IV lymph nodes as well as visualization of all possible mucosal sites of a synchronous primary. A neck CT should be performed to evaluate for cervical lymph node involvement.
Recent studies have examined the effectiveness of 18fluorodeoxyglucose positron-emission tomography (FDG-PET) in the staging of esophageal cancer. Following standard staging for esophageal cancer (including CT and endoscopy) undetected metastatic disease was detected by PET in 15% of patients in the series by Flamen et al.26 and 20% in the series by Downey and associates.27 Others have confirmed the utility of PET in staging, response to preoperative therapy, and prognosis.28–30 and should be included in the standard work-up.
Staging
The most accurate staging is pathologic.31 The American Joint Commission on Cancer (AJCC) and the Union Internationale Contre le Cancer (UICC) TNM staging systems are similar. Although the length of the primary tumor (greater than or less than 5 cm) is prognostic, the 5th edition of the AJCC staging system no longer includes length (Table 36-2). Similar to the staging system used for gastric and colorectal cancers, the T stage is based on the extension of the primary tumor through the wall.
Since many patients are treated with combined modality therapy (CMT) alone or preoperatively, clinical staging, although less accurate, may understage patients. Rizk et al. report that the AJCC staging system does not accurately predict survival in patients receiving neoadjuvant therapy.32
One of the most controversial areas is the delineation between adenocarcinomas of the GE junction and the stomach. These cancers commonly involve both the esophagus and stomach. There is no standard approach how to define the organ of origin. The most common (and practical) definition is based on where the majority of the tumor is. Whichever organ has the largest percentage of tumor is considered the primary site. Siewert and colleagues proposed a topographic anatomic classification system to differentiate three types of tumors arising in this region. They distinguish an adenocarcinoma of the distal esophagus (AEG Type I) which may infiltrate the GE junction from above from an adenocarcinoma of the gastric cardia (AEG Type II) and both of these are distinct from a subcardial gastric carcinoma the infiltrates the GE junction from below (AEG Type III).33
Standard Therapeutic Approaches
There is considerable controversy as to the ideal therapeutic approach. In the 1996 to 1999 U.S. Patterns of Care Study, 414 patients who received radiation therapy as part of definitive or adjuvant management at 59 institutions were surveyed.34,35 Overall, 51% had adenocarcinoma and 49% squamous cell carcinoma. With a median follow-up of 8 months, multivariate analysis revealed that patients who received CMT followed by surgery had a significant decrease in locoregional recurrence (HR: 0.40, P < .0001) and survival (HR: 0.32, P < .001) compared with those who did not undergo surgery. A similar significant decrease in locoregional recurrence (HR: 1.36, P = .01) and survival (HR: 1.32, P < .03) was seen in those patients who received their care at large radiation oncology centers (those that treat ≥500 new cancer patients per year) compared with small centers (those that treat <500 new cancer patients per year). In a similar pattern of care study of 767 patients treated in Japan from 1998 to 2001, 220 (28%) received preoperative or postoperative radiation or both, with or without chemotherapy.35 The National Comprehensive Cancer Network (NCCN) offers general guidelines for therapy.36
Techniques of Radiation Therapy
General Techniques
Although CT can identify adjacent organs and structures, it may be limited in defining the extent of the primary tumor. To assess the consistency of target volume delineation, Tai and colleagues sent sample cases with CT scans to 48 radiation oncologists throughout Canada and asked them to complete questionnaires regarding treatment techniques as well as outline the boost target volumes.37 There was substantial inconsistency in defining the planning target volume, both in the transverse and longitudinal dimensions. The integration of other imaging modalities in radiation treatment planning such as esophageal ultrasound, PET scan, and magnetic resonance imaging (MRI) are under active investigation.
Leong and colleagues from the Peter MacCallum Cancer Center have demonstrated that the addition of PET/CT information for treatment planning improved the identification of the gross tumor volume (GTV).38 The GTV based on CT information alone excluded PET-avid disease in 11 of 16 patients (69%), five of whom would have had a geographic miss of the gross tumor.
The standard radiation dose for patients selected for CMT is 50.4 Gy at 1.8 Gy per fraction.39 Randomized data from France reveal a higher local control (57% versus 29%) and 2-year survival rate (37% versus 23%) with continuous-course compared with split-course radiation.40 The radiation field should include the primary tumor with 5-cm superior and inferior margins and 2-cm lateral margins. The primary local/regional lymph nodes should receive the same dose. For cervical (proximal) primary tumors (defined as at or proximal to the carina) the treatment volume includes the bilateral supraclavicular nodes and for GE junction (distal) primaries the celiac axis nodes need to be included.
Treatment by Primary Site
For cervical primaries, patients are placed supine. Various field designs are possible and their choice depends on the geometry of the primary tumor in relation to the spinal cord. The ideal design is a three-field technique (two anterior obliques and a posterior). However, since the primary tumor is rarely limited to the midline, the most common approach is anteroposterior (AP)/posteroanterior (PA) to 39.6 to 41.4 Gy (Fig. 36-3) followed by a left or right opposed oblique pair with photons to 50.4 Gy (Fig. 36-4A-C). Since this technique will exclude the ipsilateral supraclavicular fossa, a separate electron field is added (commonly to a depth of 2 to 3 cm depending on the patient’s anatomy) thereby bringing the total dose to 50.4 Gy.
FIGURE 36-3 • Contours for a proximal esophageal cancer. Planning target volume (PTV) (blue) and gross tumor volume (red).
For distal (GE junction) primaries, patients are treated supine using the same four-field technique. With a CT simulation, sometimes it is difficult to distinguish normal stomach from the primary tumor. In this setting, the extent of tumor involvement on the staging PET/CT scan as well as the endoscopy report may help to define the caudal extent of the tumor. The greater curvature of the stomach is commonly used for the anastomosis. Therefore, for patients receiving preoperative treatment, care should be taken to exclude as much of the normal stomach as possible. An example of this technique is seen in Fig. 36-5A-B.
Investigators at the M.D. Anderson performed retrospective treatment planning studies on 10 patients and found that IMRT decreased the dose-volume of exposed normal lung but had no clinically meaningful differences on the irradiated volumes of spinal cord, heart, liver, or total body integral doses.41 The impact of respiratory and organ movement on defining target volumes is just starting to be identified.42 Hong et al. provide a comprehensive review of technical considerations in treatment planning.43 Examples are seen in Fig. 36-6A-C.
Critical Normal Tissues
Acute and subacute toxicity due to thoracic radiotherapy are clearly linked to the dose and volume of lung treated. However, the acceptable level of dose to the lung in the combined modality setting has not been defined. Various single parameters have been proposed to estimate the probability of developing radiation pneumonitis after radiotherapy 44–52 Investigators from the Netherlands47 compared different normal tissue complication probability (NTCP) models to predict radiation pneumonitis. Using the observed incidence of radiation pneumonitis among breast cancer, malignant lymphoma, and inoperable non–small cell lung cancer (NSCLC) patients, they found that the underlying local dose–effect relation for radiation pneumonitis was linear. This was better represented by the mean lung dose (MLD) model than by a step-function model represented by a threshold dose such as the volume receiving 20 Gy (V20). In their patient population, the MLD was the most accurate predictor for the incidence of radiation pneumonitis. Willner and colleagues performed an analysis of pneumonitis risk from DVH parameters among patients treated with three-dimensional conformal radiotherapy (3DCRT).45 Their data indicated that it is reasonable to disperse the dose outside the target volume over large areas in order to reduce the volumes of lung receiving >40 Gy. They found that reducing the high-dose volume reduces the pneumonitis rate more than a corresponding reduction in the low-dose regions of the DVH. Investigators from Memorial Sloan-Kettering Cancer Center found that grade 3+ radiation pneumonitis was significantly correlated with MLD and Fdam (an estimate of the fractional volume of lung damaged during treatment) for patients treated to higher doses (range: 57.6 Gy to 81 Gy) for NSCLC.46 Graham et al. reported that among patients treated with radiation therapy alone for NSCLC, the risk of grade 2+ pneumonitis was 0% when the V20 was <22%, 7% when the V20 was 22% to 31%, 13% when the V20 was 32% to 40%, and 36% when V20 was >40%.50 Table 36-3 summarizes the radiation pneumonitis risks associated with dosimetric criteria, V20 and MLD in several studies. Based on these studies and others, the Radiation Therapy Oncology Group (RTOG) adopted the cut-offs of a V20 ≤37% and MLD ≤20 Gy to identify patients at high risk for radiation pneumonitis in studies of NSCLC.
Most of these parameters are derived from studies of definitive radiation CMT rather than pre- or postoperative treatment. Postoperative pulmonary complications may be an even more relevant endpoint for patients with esophageal cancer, because the impact of radiotherapy on lung function may be greater in patients who undergo a thoracotomy and further insult to the lungs during surgical resection. One recent study showed that dosimetric differences impact clinical outcomes in the combined modality setting, with preoperative lung dose an important predictor of postoperative pulmonary complications. In the treatment of esophageal cancer using a neoadjuvant regimen of 45 Gy with concurrent chemotherapy, a lung V10 of 40% or greater, and a V15 of 30% or greater, was shown to be predictive of significantly greater pulmonary complications.53 Several other publications have evaluated the effect of low dose radiotherapy on the risk of pneumonitis and postoperative pulmonary complications. It appears likely that normal tissue tolerance in neoadjuvant therapy (i.e., preoperative CMT followed by surgical resection) may differ from that seen with definitive radiation alone, and even relatively low doses when delivered to a sufficient percentage of lung volume may be less tolerable when given preoperatively with concurrent chemotherapy.
Outcome
Surgery
Surgery is one of the two standard therapies for clinically resectable esophageal cancer. Two randomized trials of preoperative chemotherapy (fluorouracil [5-FU]/cisplatin over two cycles) have been performed. The INT 0113 trial revealed no survival advantage.54,55 Although the Medical Research Council (MRC) trial did show a 10% survival advantage, 9% of patients received chemotherapy plus radiation.56 In the surgical control arm from INT 0113, the median survival was 15 months and the 5-year survival was 20%. Major surgical controversies include the operative approach (transthoracic or Ivor-Lewis esophagectomy versus transhiatial esophagectomy) and the extent of resection. Because of high complication rates, three-field dissection is rarely performed in Western countries.57 Likewise, a minority of surgeons from Western countries advocate en bloc resections. Although some phase II trials suggest improved survival with an en bloc resection, this may be due to selection bias, since this operation allows more-pathologically-accurate staging. In a recent series of 111 patients (of which 10% also had preoperative adjuvant therapy), the 5-year survival of the 67 patients with node-positive disease was 26%, which is similar to the surgical control arm of INT 0113.58
Nonsurgical Therapy
Radiation Therapy Alone
Many historical series have reported results of external beam radiation therapy (EBRT) alone. Most include patients with unfavorable features, such as clinical T4 disease. Overall, the 5-year survival rate for patients treated with conventional doses of radiation therapy alone is 0% to 10%.59–61 Shi and colleagues reported a 33% 5-year survival rate with the use of late course accelerated fractionation to a total dose of 68.4 Gy.62 However, in the “radiation therapy alone” arm of the RTOG 85-01 trial, in which patients received 64 Gy at 2 Gy/day with conventional techniques, all patients were dead of disease by 3 years.63,64 There is limited experience with the use of radiation in the curative setting for patients with cT1N0 disease. Sai and colleagues from Kyoto University treated 34 patients who were either medically inoperable or refused surgery with either EBRT alone (64 Gy) or EBRT (52 Gy) plus 8 to 12 Gy with brachytherapy.65