Adult Chest Surgery

Chapter 18. Three-Field Esophagectomy


The esophagus traverses the neck, mediastinum, and abdomen. Cancer of the thoracic esophagus can metastasize to lymph nodes and locate in any or all of these compartments.The rate of lymph node involvement is very high. Approximately 50% of tumors that invade the submucosa develop lymph node metastases, and the rate increases with increasing depth of invasion. Radical surgery for esophageal cancer therefore requires three-field lymphadenectomy. Our group has been performing three-field lymph node dissections since 1984 for all thoracoesophageal cancers. In this chapter we lay down the principles, describe the procedure, and discuss the outcome of this mode of treatment based on our experience.

The overall 5-year survival rate for squamous cell carcinoma of the esophagus has improved from 20% to 50%. We believe this improvement in survival can be directly related to extensive and meticulous lymphadenectomy.This view has been corroborated by multivariate analysis. The key to three-field lymphadenectomy therefore is meticulous dissection of the upper mediastinum and cervical nodes that lie along the course of both recurrent laryngeal nerves. Preserving the right bronchial artery and pulmonary branches of the vagus nerve decreases the rate of pulmonary complications. Using this comprehensive approach, we have achieved a postoperative mortality rate of less than 2%.


Esophageal cancer is more aggressive biologically than other gastrointestinal malignancies and has a higher incidence of lymph node metastasis.Lymph node metastasis is an important and independent factor for predicting the prognosis of esophageal cancer. The number of metastatic lymph nodes is thought to reflect the aggressiveness of the cancer.Accurate documentation of the extent of lymph node involvement therefore is essential to determining the appropriate treatment strategy for esophageal cancer.

Histopathologic assessment remains the gold standard for accurate lymph node staging. Proper assessment also requires that an adequate number of lymph nodes be presented to the pathologist. Since 1984, all tumors determined to have invasion of the submucosa or beyond undergo three-field dissection at our institution. Three-field dissection is defined as an extended en-bloc lymph node dissection throughout the cervical, thoracic, and abdominal fields.5


A total of 251 patients underwent transthoracic esophagectomy with extended en-bloc cervicothoracoabdominal (three-field) lymphadenectomy between January 1998 and December 2002 at Juntendo University in Japan. A total of 27,774 lymph nodes were dissected, and the average number of dissected lymph nodes per patient was 111. The lymph nodes removed en bloc with the specimen were dissected and classified into respective lymph node groups immediately after the operation by the surgeons who performed the esophagectomy, as outlined in the Japanese Guidelines for Clinical and Pathologic Studies on Classification of Esophageal Cancer(Fig. 18-1). This provides a more detailed lymph node classification than the AJCC Cancer Staging Manual.The pattern of lymphatic spread was investigated in detail, and the final pathologic diagnosis of lymph node metastasis was compared with the preoperative clinical evaluation to assess the accuracy of preoperative diagnosis for each lymph node station and each field.

Figure 18-1.


Precise classification and station numbers of regional lymph nodes according to the Japanese Classification of Esophageal Cancer. (Illustration adapted from an original sketch provided by author.)


Rate of Lymph Node Metastasis

Seventy-three patients did not have any lymph node metastases, whereas 178 patients had one or more metastatic lymph nodes, yielding a metastatic rate of 70.9%. The rate of lymph node metastasis increased with the depth of tumor invasion and was 58.6% for pT1b, 73.5% for pT2, and 82.0% for pT3 or pT4 disease (Fig. 18-2). We divide pT1 tumors into two subclasses, pT1a and pT1b. A pT1a tumor invades only mucosa, including muscularis mucosae (mucosal cancer), and pT1b tumor invades the submucosal layer (submucosal cancer).

Figure 18-2.


Rate of lymph node metastasis according to tumor invasion. Comparison between esophageal and gastric cancer.


As shown in Fig. 18-2, the rate of lymph node metastasis in esophageal cancer is three times higher than that of gastric cancer even for submucosal invasion (pT1b). The mean number of dissected lymph nodes was 42 in the neck, 37 in the mediastinum, and 32 in the abdomen. The mean number of metastatic lymph nodes was 1.2 in the neck, 2.0 in the mediastinum, and 1.5 in the abdomen.

The frequency and distribution of lymph node metastases differ according to the location of the tumor. The rate of lymph node metastasis was 54.3% in upper thoracic esophagus (n = 35), 74.3% in middle thoracic esophagus (n = 152), and 73.4% in lower thoracic esophagus (n = 64) (Fig. 18-3). Upper esophageal tumors had a greater frequency of metastases to cervical lymph nodes than tumors of middle or lower esophagus. The frequency of metastasis to abdominal nodes was higher with lower esophageal cancer than with tumors of the middle and upper esophagus. However, we did sometimes find abdominal lymph node metastasis in upper esophageal cancer and cervical node metastasis in lower esophageal cancer. Midesophageal tumors frequently metastasized to lymph nodes in the neck and abdomen. Approximately 90% of cervical node metastases were caudal to the superior belly of the omohyoid muscle (Fig. 18-4).

Figure 18-3.


Frequency and distribution of lymphatic spread according to location of esophageal cancer. (Illustration adapted from an original sketch provided by author.)


Figure 18-4.


Lymph node metastasis in the neck. Among 140 patients with cervical metastatic nodes, 122 patients (87.1%) had positive nodes caudal to the omohyoid muscle. (Illustration adapted from an original sketch provided by author.)

Which Are the Frequently Involved Lymph Node Groups?

Analysis of lymphatic spread revealed that the sites with a metastatic rate of more than 20% (main metastatic sites) were located along the right recurrent laryngeal nerve in the upper mediastinum (106-recR: 22.2%) and along the lesser curvature of the proximal stomach (3: 23.6%) (Fig. 18-5). The sites with a metastatic rate ranging from 10% to 20% (common metastatic sites) were the right supraclavicular area (104R: 10.0%), right cervical paraesophageal area along the recurrent laryngeal nerve (101R: 11.6%), along the left recurrent laryngeal nerve in the upper mediastinum (106-recL: 19.6%), the subcarinal area (107: 12.8%), the middle thoracic paraesophageal area (108: 16.0%), the lower thoracic paraesophageal area (110: 13.8%), the posterior mediastinal area (112: 10.8%), along the thoracic duct (TD: 13.8%), the right (1: 14.3%) and left (2: 15.2%) pericardial area, and along the left gastric artery (7: 19.7%). The nodes involved less often by metastases (metastatic rate < 5%) were the middle deep cervical nodes (102R: 2.8%, 102L: 4.0%), the diaphragmatic nodes (111: 0.8%), the nodes along the common hepatic artery (8: 3.8%), and the nodes along the splenic artery (11: 4.5%).

Figure 18-5.


Main and common metastatic regional lymph node stations. (Illustration adapted from an original sketch provided by author.)

As noted earlier, this detailed analysis of the lymphatic spread of esophageal cancer based on pathologic findings revealed two main metastatic sites: one along the recurrent laryngeal nerve in the upper mediastinum, which Haagensen called the recurrent nerve chain,and the other along the lesser curvature of the proximal stomach. Unlike other gastrointestinal malignancies, frequent metastases to distant nodes are a distinctive feature of esophageal cancer.


The preoperative diagnostic workup for lymph node metastasis consists of endoscopic ultrasonography, conventional ultrasonography of the neck and abdomen, and CT scan from the neck to the lower abdomen. A PET scan is done as indicated.

Criteria for Preoperative Diagnosis

Mediastinal and abdominal lymph nodes were considered to be metastatic when the largest diameter was greater than 10 mm, the node was almost round, the internal CT density or ultrasound echogenicity was low, and the margin of the node was clear. For cervical lymph nodes, the same criteria were applied, except that the diameter was set at greater than 5 mm. Figure 18-6 shows a metastatic lymph node along the right recurrent nerve in the superior mediastinum at the root of the right subclavian artery.

Figure 18-6.


The arrow shows a metastatic node along the right recurrent laryngeal nerve (106-recR).

Accuracy of Preoperative Diagnosis

Figure 18-7 depicts the accuracy of preoperative diagnosis of lymph node metastases compared with the subsequent pathologic examination. For each lymph node station, the true-positive rate was 85–94% in the neck, 79–86% in the thorax, and 78–86% in the abdomen. The false-negative and false-positive rates for metastatic lymph nodes are also shown in the figure.

Figure 18-7.


True-positive, false-negative, and false-positive clinical diagnostic rates for lymph nodes metastases at each regional lymph node station. A. Cervical field. B. Thoracic field. C. Abdominal field. (Illustration adapted from an original sketch provided by author.)

Stringent assessment recognizes diagnostic failure at any lymph node station as a misdiagnosis for the patient. According to this criterion, the true-positive rate for lymphatic spread was 71.3% in the neck, 42.0% in the mediastinum, and 57.9% in the abdomen. The true-positive rate for all lymph node stations throughout the three fields was reduced to 23.2%. Likewise, the false-negative rate for lymph nodes metastases was 13.9% in the cervical field, 35.6% in the thoracic field, and 33.5% in the abdominal field. The overall false-negative rate for lymph node metastases increases to 53.7% across all three fields.


Radical esophagectomy should encompass all lymph node stations having a greater than 10% incidence of metastases. Nevertheless, lymphatic metastasis cannot be diagnosed precisely either by ultrasonography or CT imaging before surgery. Therefore, radical surgery for cancer of the thoracic esophagus requires complete three-field lymph node dissection.

The first step of the operation is an anterolateral thoracotomy through the fourth intercostal space. The thoracic duct, azygos vein, pleura, and periesophageal tissues including lymph nodes and lymphatic channels in the mediastinum are dissected en bloc. The right and left recurrent laryngeal nerves are identified, and the upper mediastinal lymph nodes (including the node group of the recurrent laryngeal nerve chain) are cleared (Fig. 18-8). After this procedure, paratracheal lymph nodes on both sides (Fig. 18-9), subcarinal, right and left hilar lymph nodes, posterior mediastinal lymph nodes adjacent to the descending aorta and left pleura, and diaphragmatic lymph nodes are dissected en bloc (see Fig. 18-1). The lymph nodes inside the aortic arch (left tracheobronchial) are dissected separately (Fig. 18-10). To avoid ventilatory impairment, the right or left bronchial artery and the pulmonary branches of the vagus nerve should be preserved (Fig. 18-11).

Figure 18-8.


Lymph node dissection is started with identification of the right recurrent laryngeal nerve in the utmost upper mediastinum. (Illustration adapted from an original sketch provided by author.)


Figure 18-9.


Upper mediastinal dissection. The right and left recurrent nerves are separated, and tissues are cleared around the nerves. (Illustration adapted from an original sketch provided by author.)


Figure 18-10.


Completion of dissection inside the aortic arch. (Illustration adapted from an original sketch provided by author.)


Figure 18-11.


Preservation of the right bronchial artery and pulmonary branches of the vagus nerve. (Illustration adapted from an original sketch provided by author.)


For the abdominal procedure, after an upper midline laparotomy, en-bloc dissection of lymph nodes is carried out along the cardia, lesser curvature, left gastric artery, celiac axis, common hepatic artery, and splenic artery (Fig. 18-12). The left gastric artery is cut at its origin. Along with all these dissected lymph nodes, the proximal stomach is cut between the junction of the right and left gastric arteries to the highest point in the fundus using a linear stapler (Fig. 18-13). The gastric remnant, based on the right gastroepiploic artery and right gastric artery, is used for esophageal reconstruction. We do not perform pyloroplasty. The extent of abdominal lymph node dissection is very similar to that of D2 lymph node dissection for gastric cancer surgery.

Figure 18-12.


Dissection areas in the abdomen. (Illustration adapted from an original sketch provided by author.)


Figure 18-13.


The cut line of the stomach. The proximal stomach is cut between the junction of the right and left gastric arteries to the farthest point on the fundus. (Illustration adapted from an original sketch provided by author.)


In the neck, a collar incision is made as in thyroidectomy. The middle deep cervical and supraclavicular lymph nodes (102 and 104 in Fig. 18-1), which are located lateral to the common carotid artery, superior to the anterior scalene muscle and phrenic nerve, and inferior to the superior belly of the omohyoid muscle, are removed. Then the lymph nodes along the cervical recurrent laryngeal nerve are excised (Fig. 18-14). These lymph nodes are between the common carotid artery and trachea (101 in Fig. 18-1), and they should be removed meticulously and carefully so as not to traumatize the recurrent laryngeal nerve.

Figure 18-14.


Dissection along the cervical recurrent laryngeal nerve on the left side. (Illustration adapted from an original sketch provided by author.)


Postoperative Morbidity

The patient is usually extubated in the operating theater when the PaO2 level (torrs) exceeds three times the FIO2 (%). The patient is transferred to the surgical ICU and is monitored there for a period of approximately 1 week.

The surgeon should be aware of four main complications in the postoperative course of esophagectomy with three-field lymph node dissection (Fig. 18-15). They consist of pulmonary complications, cardiac complications, anastomotic leakage, and recurrent nerve paralysis. Among these, pulmonary complications are by far the most common and of grave concern. Pulmonary compromise may be caused by the wide dissection around the trachea and bronchi, which leads to various degrees of ischemia of the respiratory tract and a decreased cough reflex.

Figure 18-15.


Postoperative complications after three-field dissection. Anastomotic leakage includes radiologic minor leakage (2.8%) and clinical fistula formation (3.2%).


(See Fig. 18-16.) Hypoxemia requires inhalation of a high concentration of oxygen and was seen in 25.4% of the study group. It is caused mainly by pneumonia, atelectasis, and pulmonary edema. Pulmonary edema or potential pulmonary edema sometimes develops in radical esophagectomy patients after removal of the thoracic duct, which causes lymphatic retention in the retroperitoneum, hypoproteinemia, and depletion of intravascular volume. The rate of fluid infusion during the procedure is maintained at approximately 7–8 mL/kg/h, taking into account the blood loss and urine output (Fig. 18-17). Consequently, potential pulmonary edema can occur in the resorption period approximately 48 hours postoperatively. It can be controlled successfully by administration of dopamine, albumin, and diuretics (e.g., furosemide). It should be borne in mind that intraoperative restriction of fluid transfusion to keep the lungs "dry" can result in tachycardia with serious hypotension postoperatively.

Figure 18-16.


Pulmonary complications after three-field dissection.


Figure 18-17.


Perioperative management of three-field dissection.

Pulmonary embolism is a less common complication, but one should maintain a high degree of suspicion in patients with no obvious cause of hypoxemia. Its prevention is very important. It is advisable to use pneumatic pressure garments on both legs and to administer heparin (200–250 U/kg/d) throughout the perioperative period until patients are mobile.


Cardiac complications occurred in 22.9% of patients. They mainly consist of tachycardia, arrhythmia, atrial fibrillation, premature ventricular contractions, and so forth. These complications are usually seen in the resumption period as a consequence of cardiac overload because of resorption of fluid from the third space. Cardiac overload can be controlled by administration of diuretics. Tachycardia also may occur because of intravascular hypovolemia. Differentiating overhydration from underperfusion is important to implement proper corrective measures. Low-dose dopamine (2 g/kg/min) should be administered continuously for approximately 1 week postoperatively to ensure better circulation and maintain diuresis.


We use a hand-sewn two-layer anastomosis. The inner-layer suture is continuous and consists of a 5-0 monofilament absorbable material. The outer layer is completed with interrupted fine silk sutures. Clinical leakage involving fistula formation occurred in 3.2%. All patients underwent a barium swallow on the ninth postoperative day, and 6.0% of the patients were diagnosed as having anastomotic leakage. Minor leakage usually healed within 1–2 weeks of conservative management with total parenteral nutrition.


Recurrent laryngeal nerve palsy is one of the causes of pulmonary complications. In our series, 10.1% of patients with three-field lymph node dissection suffered from recurrent laryngeal nerve palsy, and 95% of them had impairment of the left recurrent nerve. This may cause silent aspiration of saliva, which contains bacteria from the oral cavity. Inadequate glottic closure also impairs the cough reflex and expulsion of retained secretions by preventing buildup of adequate intratracheal pressure. Hoarseness owing to unilateral nerve palsy recovers within 3 months, but bilateral nerve palsy occasionally may require tracheostomy to bypass the closed glottic aperture caused by the unopposed action of the cricothyroid muscle.

Postoperative Mortality

Mortality within 30 days of surgery is called direct operative death. Mortality of patients who succumbed for any reason during hospitalization even after the 30-day period is called postoperative hospital death, and this includes direct operative death. In our series from January 1998 to October 2005, 436 patients underwent esophagectomy with three-field lymph node dissection and reconstruction. We had 5 direct operative deaths and 13 postoperative hospital deaths, including the 5 direct operative deaths. The direct operative death rate was 1.07%, and the postoperative hospital death rate was 2.98%. These rates compare favorably with the mortality rates reported in other large series.


The long-term survival rate was calculated according to the Kaplan-Meier method, and the statistical difference was considered to be significant at a p value of less than 0.05.

Statistical Analysis

The survival curve of the three-field dissection group was compared with the survival curve of the historical group (limited two-field dissection) before 1984. The overall 5-year survival rate for the three-field group was 50.8%, whereas in the limited two-field group it was 37.1% (p < 0.001) (Fig. 18-18). The effect of extensive lymph node dissection on long-term survival also was examined using multivariate analysis (Cox regression model). Ten factors such as age, gender, tumor location, number of metastatic nodes, extent of lymph node dissection (three versus two fields), cellular differentiation, curativity, pT category, pN category, and clinical M category were entered into the model as covariates. Consequently, the extent of lymph node dissection was selected as one of the independent prognostic factors with a risk ratio of 0.5501 (Table 18-1). This means that systematic lymph node dissection contributes to improved long-term survival for squamous cell carcinoma of the thoracic esophagus.

Figure 18-18.


Comparison of survival curves between two- and three-field dissection (Kaplan-Meier method).

Table 18-1. Independent Prognostic Factors Examined by Cox Regression Model



Risk ratio




No. of positive node



Cell differentiation









Gender(f: m)



Residual tumor





































Relationship between the Pattern of Lymph Node Metastasis and Prognosis

Survival rate decreased as the number of metastatic lymph nodes increased irrespective of the site of metastasis. The prognosis of 386 patients who underwent three-field lymph node dissection without preoperative neoadjuvant chemo- or radiotherapy was analyzed to correlate lymph node metastasis pattern with prognosis.

The number of positive nodes has a close relationship with prognosis (Fig. 18-19). The 5-year survival rate for patients without lymph node metastasis and those with less than five metastatic nodes was 78.3% and 48.6%, respectively. When there were more than five metastatic nodes, the 5-year survival rate decreased to less than 20%. Therefore, the number of lymph node metastases is an independent predictor of prognosis. The number of fields that harbor metastatic nodes is also a factor in predicting prognosis (Fig. 18-20). Patients without metastasis showed the best survival rate of 78.3%, followed by one-field metastases and then two-field metastases. Patients with metastases in all the three fields had the worst prognosis, with a 5-year survival rate of only 14.3%.

Figure 18-19.


Survival curves according to the number of positive nodes.


Figure 18-20.


Survival curves according to metastatic fields. c,C = cervical node, t,T = thoracic node, a,A = abdominal node. Capital letter means positive node, and small letter means negative node.


In Japan, three-field lymph node dissection is used routinely for the surgical treatment of squamous cell carcinoma of the thoracic esophagus. However, this procedure has not been universally accepted. Doubts continue to persist regarding the need for such extensive dissection and the subsequent benefits conferred.We analyzed the extent and frequency of lymph node metastasis of squamous cell carcinoma of the esophagus and its impact on long-term survival. We found that the lymph node stations involved depend on the site of the primary tumor. Positive nodes were found with equal frequency in all three fields. It was thought previously that metastases to neck nodes were distant metastases, and they were considered to indicate incurability. In our experience, however, some patients with metastatic neck nodes can have prolonged survival with radical three-field dissection. Extended lymph node dissection of the neck, mediastinum, and abdomen should be considered mandatory for complete eradication of carcinoma of the thoracic esophagus.


Dr. Akiyama's group has had extensive experience with the three-field technique and has shown superior survival over two-field resections.



1. Akiyama H: Squamous cell carcinoma of the thoracic esophagus. In Akiyama H (ed): Surgery for Cancer of the Esophagus. Baltimore, Williams & Wilkins, 1990.

2. Udagawa H, Tsutsumi K, Kinoshita Y: Surgical treatment of squamous cell cancer of the esophagus. In Pinotti HW (ed): Recent Advances in Diseases of the Esophagus. Bologna, Monduzzi Editore, 2001:585–588.

3. Kuwano H, Nakajima M, Miyazaki T, Kato H: Distinctive clinicopathological characteristics in esophageal squamous cell carcinoma. Ann Thorac Cardiovasc Surg 9:6–13, 2003. [PubMed: 12667124]

4. Hosch SB, Stoecklein NH, Pichlmeier U, et al: Esophageal cancer: The mode of lymphatic tumor cell spread and its prognostic significance. J Clin Oncol 19:1970–5, 2001. [PubMed: 11283129]

5. Akiyama H, Tsurumaru M, Udagawa H, Kajiyama Y: Radical lymph node dissection for cancer of the thoracic esophagus. Ann Surg 220:364–72; discussion 72–3, 1994. 

6. Japanese Society for Esophageal Diseases: Japanese Classification of Esophageal Cancer. Tokyo, Kanehara, 2008.

7. Fleming I, Cooper J, Henson D: AJCC Cancer Staging Manual. Philadelphia, Lippincott, 1997.

8. Haagensen C, Feind C, Herter F, et al: The Lymphatics in Cancer. Philadelphia, Saunders, 1972:78–9.

9. Watanabe H, Kato H, Tachimori Y: Significance of extended systemic lymph node dissection for thoracic esophageal carcinoma in Japan. Recent Results Cancer Res 155:123–33, 2000. [PubMed: 10693246]

If you find an error or have any questions, please email us at Thank you!