19: Radical En Bloc Esophagectomy

Despite improvements in perioperative care, surgical techniques, and neoadjuvant therapy over the last decade, the prognosis of esophageal cancer remains poor. More than 95% of new cases diagnosed annually in the United States succumb to disease. Among the subset of patients resected with curative intent (R0 resection), the 5-year survival after transthoracic esophagectomy or transhiatal esophagectomy rarely exceeds 30% based on reports from large surgical series.^1-4 The principal justification for these poor results is the finding that most patients develop metastatic disease and already may have disseminated disease at the time of diagnosis. A careful analysis of the patterns of failure after surgical resection also implicates inadequate locoregional control. The locoregional failure rates are unacceptably high after conventional surgical resection, ranging from 30% to 60%.^5-8 The addition of preoperative therapy of any kind does not meaningfully reduce the high rate of local failure.^6-8 Thus a meaningful improvement in the survival of patients with esophageal cancer is unlikely without adequate locoregional control.

En bloc resection for tumor of the lower esophagus and cardia was first described by Logan in 1963.⁹ The reported 5-year survival was unparalleled at the time but was achieved at the cost of high operative mortality. In 1979, Skinner revisited the en bloc approach and extended its use to tumors of the middle and proximal esophagus, publishing his first report in 1983.¹⁰ A few years earlier, Orringer and Sloan¹¹ published their first report on the transhiatal approach for esophagectomy without thoracotomy. The controversy continues to the present concerning the efficacy of radical en bloc esophagectomy, and most surgeons favor conventional techniques of esophageal resection through either a transthoracic or a transhiatal approach. However, we and others continue to advocate radical en bloc esophageal resection as the optimal procedure for maximizing locoregional control and improving long-term survival in patients with esophageal cancer.¹² The basic concept of en bloc esophagectomy is resection of the tumor-bearing esophagus with a wide margin of surrounding tissues. Thus, for tumors of the middle or lower thoracic esophagus, the en bloc specimen includes the tumor-bearing esophagus, the pericardium anteriorly, both pleural surfaces laterally, and the thoracic duct and all other lymphoareolar tissue wedged posteriorly between the esophagus and the spine. The associated lymphadenectomy includes en bloc resection of all nodal groups in the middle and lower mediastinum as well as the upper abdomen.

For a selected group of patients, the lymphadenectomy is extended to include the superior mediastinal and cervical lymph nodes (three-field lymph node dissection). The three-field concept was first introduced by Japanese surgeons, prompted by their observation that up to 40% of patients resected by radical two-field esophagectomy developed isolated recurrences in the cervical nodes.¹³ In 1991, Isono et al.¹⁴ reported nationwide results of three-field lymph node dissection and found that occult cervical node metastases occurred in one-third of patients. Even for lower-third tumors, up to 20% of patients harbored cervical metastases. Most Western surgeons have been reluctant to adopt the three-field dissection technique for two principal reasons: skepticism that long-term survival can be achieved once nodal disease is present and the reported high morbidity associated with the operation. In particular, injury to one or both recurrent laryngeal nerves has been described in as many as 50% of patients with a consequent high risk for tracheostomy.^15,16

Preoperative assessment is directed toward establishing, as accurately as possible, the clinical stage of the disease, as well as assessing the patient's ability to tolerate the planned operation. Our standard diagnostic and staging workup includes an upper endoscopy with biopsy and CT scan of the chest and upper abdomen in all patients. In addition, patients currently undergo endoscopic ultrasonography as well as positron emission tomography. The former is useful in selecting patients for clinical trials of preoperative induction therapy, whereas the latter is a more sensitive test for detecting distant visceral and skeletal metastases. Generally, patients are considered for primary surgical resection if the preoperative evaluation reveals no evidence of distant visceral metastases or clear evidence of direct neoplastic invasion of the airway or major vascular structures. The presence of extensive nodal disease is not considered a contraindication to resection unless it clearly extends beyond the proposed fields of dissection. Finally, all patients are evaluated for pulmonary and cardiac function to determine their ability to withstand the planned procedure.

The basic principle underlying en bloc esophagectomy is resection of the tumor-bearing esophagus within a wide envelope of periesophageal tissue, which includes both pleural surfaces laterally, a patch of pericardium anteriorly, and the thoracic duct posteriorly, along with the mediastinal lymph nodes from the tracheal bifurcation to the hiatus (Fig. 19-1).

An upper abdominal lymphadenectomy is also performed, including the celiac, common hepatic, left gastric, parahiatal, lesser curvature, and retroperitoneal lymph nodes. A “third field” nodal dissection can be incorporated by extending the lymphadenectomy to include the superior mediastinal and cervical lymph nodes (Fig. 19-2). The procedure is almost always carried out through three incisions: a right thoracotomy, followed by a laparotomy and collar neck incision. More recently, the thoracic and abdominal portions of the dissection are accomplished by a minimally invasive approach.

The Thorax

A right fifth interspace thoracotomy is performed regardless of the location of the tumor within the esophagus (Fig. 19-3, inset). The “first field” comprises the middle and lower mediastinum and is bound superiorly by the tracheal bifurcation, inferiorly by the esophageal hiatus, anteriorly by the hilum of the lung and pericardium, and posteriorly by the descending thoracic aorta and the spine. Dissection of the middle and lower mediastinum begins by incising the mediastinal pleura over the anterior aspect of the azygos vein from the level of the azygos arch superiorly to the aortic hiatus inferiorly. The dissection proceeds leftward anterior to the aorta and across the mediastinum to the opposite pleura, which is entered along the entire length of the incision. The thoracic duct thus is mobilized anteriorly toward the specimen and is ligated inferiorly at the aortic hiatus and superiorly as it crosses to the left side of the mediastinum (Fig. 19-3). All lymphatic channels are clipped or ligated between the thoracic duct and the spine to minimize the probability of a chylothorax. The arch of the azygos vein, but not its main trunk, is resected en bloc with the specimen. The anterior dissection is commenced by dividing the azygos vein at its caval junction and by carrying the dissection along the right main bronchus and the posterior aspect of the hilum of the right lung. The hilar and subcarinal nodes are cleared, and a patch of pericardium is resected en bloc with the tumor-bearing esophagus for all but submucosal tumors (T1) of the middle and lower thirds of the esophagus. Division of both pulmonary ligaments (left and right) completes the esophageal mobilization (Fig. 19-4). For tumors traversing the hiatus, a 1-in cuff of diaphragm is excised circumferentially en bloc with the specimen using electrocautery. The completed dissection clears all nodal tissue in the middle and lower mediastinum, including the right and left paraesophageal, parahiatal, paraaortic, subcarinal, bilateral hilar, and aortopulmonary lymph nodes.

Dissection of the third field begins during the thoracic portion of the procedure and is later completed through a collar neck incision. Dissection of the nodes in the superior mediastinum includes the nodes along the right and left recurrent laryngeal nerves throughout their mediastinal course. The paratracheal retrocaval compartment is not disturbed. The left recurrent nerve is dissected using a “no touch” technique, and nodes along its anterior aspect are carefully excised. Notably, there is a paucity of nodal tissue along the left nerve in nearly all Caucasians. The right recurrent nerve is carefully exposed near its origin at the base of the right subclavian artery (Fig. 19-5). The right vagus nerve serves as a good guide to locate the right recurrent nerve. The right recurrent nodal chain begins at that level and forms a continuous package that extends through the thoracic inlet to the neck. Again, the nerve is dissected using a strict no touch technique. Through the cervical incision, the remainders of the recurrent nodes are dissected, as are the lower deep cervical nodes located posterior and lateral to the carotid sheath. Thus the third field includes a continuous anatomically inseparable chain of nodes that extends from the superior mediastinum to the lower neck. These nodes should be appropriately labeled cervicothoracic nodes rather than cervical nodes.

The Abdomen

The abdomen is entered through a midline incision (Fig. 19-6, inset). The omentum is separated from the colon in the avascular plane, and the lesser sac is entered. After dividing the short gastric vessels, the retroperitoneum is incised along the superior border of the pancreas (Fig. 19-6). The retroperitoneal lymphatic and areolar tissues are swept superiorly toward the esophageal hiatus and medially along the splenic artery to the celiac trifurcation. The left gastric artery is divided flush with its celiac origin, and the nodes along the common hepatic artery are dissected toward the specimen. This retroperitoneal dissection is bound by the dissected esophageal hiatus superiorly, the hilum of the spleen laterally, and the common hepatic artery and inferior vena cava medially. Finally, the lesser curvature and left gastric nodes are included with the specimen as the gastric tube is prepared. The omentum is resected as a separate specimen at least 1 in outside the gastroepiploic arcade.

The Neck

A generous low collar incision is performed, and subplatysmal flaps are raised inferiorly and superiorly (Fig. 19-7). The strap muscles and the medial heads of the sternocleidomastoid are divided. The esophagus (previously fully mobilized from the thorax) is retrieved from the prevertebral space. The esophagus is divided distally, and the specimen is retrieved in the abdomen. The previously dissected recurrent nerves are easy to visualize (especially the right recurrent nerve), and any residual nodal tissue is excised. Next, the nodes posterior and lateral to the carotid sheath are removed, along with the supraclavicular nodes, particularly for tumors of the middle and upper esophagus. The dissection is limited superiorly by the inferior belly of the omohyoid. Within the abdomen, the gastric tube is prepared and the specimen is removed (Fig. 19-8). Gastrointestinal continuity is restored by a cervical esophagogastrostomy. A feeding jejunostomy tube is placed for early postoperative enteral feeding.

Most patients are cared for in an ICU for 24 hours for fluid management. Currently, with improved epidural pain control and pulmonary physiotherapy, patients who undergo a two-field en bloc resection are extubated in the OR. Patients who undergo a three-field en bloc resection often require 24 hours of mechanical ventilation. Intense pulmonary hygiene is required, often with repeated bronchoscopy for the first 48 hours after extubation, because most patients have bronchorrhea, which generally resolves on the third or fourth postoperative day. Patients often demonstrate significant fluid sequestration postoperatively, with spontaneous diuresis by the third postoperative day. Aggressive physical therapy is essential for getting patients out of bed and ambulating. Enteral jejunostomy feeding is started by the fourth or fifth postoperative day. Chest tubes are removed when drainage is less than 250 mL/day. Oral intake is begun once anastomotic integrity is confirmed by a contrast study on the sixth or seventh postoperative day. Patients are discharged by the tenth postoperative day on a regular diet but often require supplemental jejunostomy feeding at night. The jejunostomy tube is usually removed 4 weeks after hospital discharge if the usual postoperative anorexia resolves and oral intake is considered adequate.

In 2001, we reported a series of 111 patients who underwent en bloc resection with either a two- or three-field dissection.¹⁷ All patients underwent en bloc esophagectomy for esophageal cancer between 1988 and 1998. The overall hospital mortality was 5.4%. This is similar to the mortality rates of conventional esophagectomy. Fifty-seven patients (51%) had an uncomplicated recovery. Complications occurred in 54 patients and were considered minor in 11 and major in 43 (including 6 postoperative deaths; 38.7%) (Table 19-1). The most common morbidity was pulmonary. Anastomotic leaks occurred in 13% of patients, and all healed with simple drainage. Recurrent nerve injuries occurred in only four patients and were unilateral in all. No patients required tracheostomy as a result of recurrent nerve injury.

Overall 5-year survival for all patients was 40%, with a median survival of 38 months (Fig. 19-9). Node-negative patients had a significantly improved 5-year survival of 75% compared with 26% in node-positive patients. More impressively, the 5-year survival for stage III patients was 39% compared with 11% after conventional transthoracic esophagectomy, as reported previously.¹⁸ This is especially important because most of the patients presenting with esophageal cancer have stage III disease. It is interesting to observe that for stage IV patients, 5-year survival was 27%. Survival also was significantly better for patients with locoregional N1 nodal metastases compared with distant M1a nodal metastases (31% vs. 21%, p = 0.03). Nonetheless, even for patients with distant nodal metastases (celiac or recurrent laryngeal nodes), there were long-term survivors. These data suggest that these patients should be classified as a subgroup of stage III rather than stage IV. Overall local recurrence rate was 8%, comparing favorably with the 31% to 45% incidence of local recurrence reported after conventional esophagectomy.

In 2004, Lerut published the results of a cohort of 174 R0 resections in patients with esophageal cancer using en bloc esophagectomy with three-field lymphadenectomy.¹² Hospital mortality was 1.2%, and morbidity was 58%. Overall 3-and 5-year survivals were 51% and 41.9%, respectively, with disease-free survival of 51.4% and 46.3%, respectively. The locoregional lymph node recurrence was impressively low at 5.2%. The 5-year survival for node-negative patients was 80.2% compared with 24.5% for node-positive patients. The prevalence of metastatic disease to the cervical nodes was high: 23% in patients with adenocarcinoma and 25% in patients with squamous cell carcinoma. The 5-year survival in patients with positive cervical nodes in middle-third carcinomas was 27.2%, leading the author to suggest that these nodes should be considered as regional (N1) rather than distant metastasis (M1b) in middle-third carcinomas.

A randomized trial comparing transthoracic en bloc esophagectomy with transhiatal resection was published by Hulscher in 2002.¹⁹ Although the difference in survival between the two groups was not statistically significant, there was a trend toward a survival benefit with en bloc resection at 5 years. Five-year overall and disease-free survival rates in the en bloc group were 39% and 39%, respectively, compared with 29% and 27% in the transhiatal group. Transthoracic en bloc esophagectomy was associated with higher morbidity than transhiatal esophagectomy, consistent with the wider extent of dissection and complexity of the resection. The authors subsequently published a subgroup analysis showing that in patients with Siewert type I tumors, en bloc resection was associated with an absolute survival advantage of 14% compared to transhiatal resection (51% vs. 37%).²⁰ Further analysis suggested that the benefit may be limited to patients with 1 to 8 positive nodes rather than those with higher nodal burden or those without nodal metastases.

Recently, we reported our series of 465 patients who had a R0 resection for esophageal cancer.²¹ A total of 328 patients (71%) had an en bloc resection (two-field lymphadenectomy in 129, three-field in 199). The remaining 137 patients (29%) had a conventional esophagectomy. Multivariate regression analysis was used to identify independent predictors of freedom from recurrence and disease-free survival. Other than good performance status and early pathologic stage, en bloc resection was a significant independent predictor of improved freedom from recurrence and disease-free survival.

Radical en bloc esophagectomy can be performed with low mortality and similar morbidity to conventional transthoracic or transhiatal esophagectomy. It provides the most thorough staging information. Locoregional recurrence rates are substantially reduced, with improved disease-free survival. En bloc resection appears to have a favorable impact on survival, especially in patients with limited nodal metastases.

Surveillance endoscopy on a 72-year-old man with a long history of Barrett esophagus showed an ulcerated lesion of the distal esophagus. Biopsy results showed squamous cell carcinoma. CT of the chest and upper abdomen showed no obvious esophageal mass or nodal metastases. Endoscopic ultrasonography demonstrated a T2N0 esophageal carcinoma. Preoperative cardiopulmonary evaluation indicated that the patient was a good candidate for surgical resection, and he underwent en bloc esophagectomy with three-field lymphadenectomy. The patient was extubated on postoperative day 1 and had an episode of atrial fibrillation on postoperative day 3 that was successfully treated pharmacologically. Barium study on postoperative day 6 revealed no evidence of a leak, and oral diet was commenced. The patient was discharged home on postoperative day 8 on supplemental jejunostomy feeding. Final pathology revealed a T1N1M0 (stage IIB) squamous cell carcinoma of the distal esophagus with a single nodal metastasis in the right recurrent laryngeal nodal chain. The patient received no adjuvant therapy. After 5 years of follow-up, the patient remains disease-free.

Clearly, this technique has not “caught on” yet among the majority of esophageal surgeons. The Cornell group is to be commended, however, as their careful reporting of the data also has resulted in their recent adaptation of this technique to minimally invasive esophagectomy. As yet, no convincing data comparing radical versus nonradical or two-field versus three-field esophagectomies have emerged to show a survival advantage stage by stage. It will be interesting to see whether there are differences going forward, especially given the changes in N that appear in the new AJCC staging system (7th edition). Other than the major risk of tracheostomy postoperatively, the procedure can be done with similar results by experienced surgeons.