Comparison of transhiatal (THE) to transthoracic (TTE) esophagectomy perhaps more accurately evaluates surgical access rather than extent of lymphadenectomy. However, as the extent of lymph node removal is clearly different between the two approaches, data from randomized trials may be used to infer the importance of lymphadenectomy. In 2002, Hulscher reported the largest randomized trial comparing THE to TTE [9]. Two hundred and twenty treatment naive patients, were randomly assigned to either THE with single field upper abdominal lymph node dissection (D2 lymphadenectomy) or TTE with two-field (D2 lymphadenectomy and middle and lower mediastinal node dissection) performed by an en-bloc technique. The total lymph nodes removed were significantly higher in the TTE approach, 31 ± 14 versus 16 ± 9 with the TTE approach (p = 0.001). Although THE was associated with lower morbidity, there was a non-significant trend for improved survival favoring the transthoracic en-bloc procedure (TTE 39 % vs. THE 29 %). A subsequent per-protocol subgroup analysis showed that patients most likely to benefit from the extended resection were those with adenocarcinoma of the distal tubular esophagus (51 % vs. 37 %; p = 0.33) and those with limited nodal involvement (1–8 positive nodes) (TTE: 64 %, THE: 25 %; p = 0.02) [10]. Patients with tumors of the gastroesophageal junction and those with either no or extensive nodal metastases (>8 nodes) had similar survival with either type of resection.

The most recent meta-analysis of TTE versus THE reviewed 5,905 patients from 52 articles, including recent literature from the 2000s [11]. Patients undergoing TTE had an average of eight more lymph nodes removed than those undergoing THE (p = 0.02, CI 1–14). However, lymph node yield was adequately reported in only four studies. There was no significant difference in 5-year overall survival between patients undergoing TTE versus THE, although significant heterogeneity existed among studies.

Logan introduced the en bloc concept in 1963, which was later re-introduced by Skinner in 1979 [12, 13]. The basic premise of the en bloc operation is to maximize local tumor control by resection of the tumor bearing esophagus within a wide envelope of adjoining tissues that includes both pleural surfaces laterally and the pericardium anteriorly where these structures are intimately related to the esophagus. In addition to providing for a greater circumferential margin, en bloc esophagectomy leads to enhanced lymphadenectomy. Posteriorly, the lymphatics wedged dorsally between the esophagus and the aorta, including the thoracic duct throughout its mediastinal course, are resected en bloc with the specimen. This posterior mediastinectomy necessarily results in a complete mediastinal node dissection from the tracheal bifurcation to the esophageal hiatus. Additionally, as part of the en bloc operation, an upper abdominal lymphadenectomy is performed including the common hepatic, celiac, left gastric, lesser curvature, parahiatal, and the retroperitoneal nodes. Local recurrence rates reported by proponents of this approach have been in the 2–10 % range, much lower than those reported following either transhiatal esophagectomy or standard transthoracic resections [1, 14].

The only randomized trial reported to date comparing transthoracic en-bloc resection with non en bloc resection (transhiatal) esophagectomy is the Hulscher trial, previously discussed [9]. Again, as differences exist in the surgical access, approach, and fields of lymph node dissection this trial should not be taken as only a direct comparison of en-bloc versus non en-bloc strategies. However, as mentioned, en bloc esophagectomy enhanced lymphadenectomy and appeared to improve survival in patients with limited nodal involvement (one to eight positive nodes) (TTE: 64 %, THE: 25 %; p = 0.02).

Several retrospective case series also exist describing enhanced lymphadenectomy and patient survival with en bloc techniques. Lee et al. have recently updated their institution’s series of resections for esophageal cancer patients, from which they had previously reported the superiority of en bloc resection to non-en bloc approaches [15]. This included 465 patients who had an R0 resection, 179 patients resected following induction therapy and 286 patients treated with surgery alone. Three hundred twenty-eight patients (71 %) had an en bloc resection. The remaining 137 patients (29 %) had a non-en bloc resection (88 transhiatal, 49 transthoracic). The number of resected lymph nodes was significantly higher for the en bloc group (median 31 vs. 17, p < 0.001). For patients with pathologic stage 0/I disease, there was no significant difference in disease free survival (DFS) between the en bloc group and the standard resection group (5 year-DFS 75.7 % (95 % CI: 62.2–90.4) vs. 76.3 % (95 % CI: 65.3–86.1). However, for patients with pathologic stage II/III/IV disease, DFS was significantly improved following en bloc resection compared to standard resection [HR: 0.66, (0.50–0.88), p = 0.004]. Median DFS was 19.0 months (95 % CI: 14.0–24.0) after en bloc and 12.2 months (95 % CI: 7.7–16.7) following standard resection. An important criticism of most of this and other retrospective studies is the failure to clearly define the criteria for patient selection for one procedure versus another. For example, in the studies by Lee et al and by Hagen et al, the patients receiving a transhiatal resection were either significantly older than the en-bloc group or had a worse performance status with respect to cardiopulmonary function [15, 16].

The concept of 3-field lymph node dissection for esophageal cancer was developed by Japanese surgeons in the 1980s in response to the observation that as many as 40 % of patients with resected squamous cell esophageal cancer developed isolated cervical lymph node metastases [17]. A nationwide retrospective study was subsequently reported describing the findings and potential benefits of esophagectomy with 3-field dissection [18]. The additional third field of dissection included excision of the nodes along both recurrent nerves as they course through the mediastinum and neck, as well as a modified cervical node dissection. Previously, unsuspected cervical nodal metastases, primarily in the recurrent nodes, were seen in approximately one third of patients. Furthermore, the authors reported a significantly higher overall 5-year survival after 3-field dissection in comparison to 2-field dissection.

The largest Japanese study from a single institution was reported by Akiyama in 1994 [19]. The authors reported their experience with 717 patients in whom a complete (R0) resection was performed using either a two-field (n = 393) or a three-field technique (n = 324). Five-year survival in node-negative patients was 84 % after the three-field procedure compared to 55 % after two-field lymphadenectomy (p = 0.004). Patients with node-positive disease also fared better after three-field dissection with a 5-year survival rate of 43 % compared to a 28 % 5-year survival rate after two-field dissection (p = 0.0008).

Two prospective studies have been reported [20, 21]. The study by Nishihira was a prospective randomized trial that showed a survival advantage for three-field over two-field lymph node dissection (65 % versus 48 %); however the difference was not statistically significant [20]. The study from the National Cancer Hospital in Tokyo was a prospective non-randomized case- matched study that showed that 5-year survival was significantly better after three-field dissection (48 % versus 33 %; p = 0.03) [21].

A recent meta-analysis of the eastern literature was published by Ye et al. including 13 studies for analysis, of which two were randomized trials [22]. Total number of lymph nodes removed with each technique was not analyzed. Among 2,379 patients, a better 5-year survival rate was demonstrated in patients undergoing three-field versus two-field lymphadenectomy (HR for three-field 0.64, CI: 0.56–0.73, p < 0.001). Patients undergoing three-field lymphadenectomy had similar rates of perioperative mortality (RR 0.64, CI: 0.38–1.10, p = 0.110) and pulmonary complications (RR 1.00, CI: 0.89–1.12, p = 0.760), but higher rates of anastomotic leakage (RR 1.46, CI 1.19–1.79, p < 0.001).

The relevance of these findings to a western population afflicted primarily by esophageal adenocarcinoma remains unknown. The early experience in North America with this technique was reported by Altorki et al, in 2002 [23]. The procedure was performed in 80 patients, 60 % of which had adenocarcinoma of the esophagus. Recurrent nerve injury occurred in 6 % of patients. An average of 60 nodes were resected per patient. The prevalence of cervical nodal metastases was 37 % in patients regardless of cell type or location of the tumor within the esophagus. Overall and disease-free survival was 50 and 46 %, respectively, and was not influenced by cell types. Patients with adenocarcinoma who had metastases to the recurrent laryngeal lymph nodes had a 3- and 5-year survival of 30 and 15 %, respectively. In contrast patients with squamous cell carcinoma and positive recurrent laryngeal nodes had a 5-year survival of 40 %.

Lerut reported the most significant European experience with esophagectomy and three-field lymph node dissection [24]. One hundred and seventy four patients had an R0 three-field esophagectomy with a hospital mortality of 1.4 % and morbidity of 57 %. Fifty five percent of patients had adenocarcinoma of the esophagus or cardia. Overall and disease-free survival at 5 years was 42 and 46 %, respectively. The incidence of positive cervical nodes in patients with adenocarcinoma was 23 % and was slightly higher for those with esophageal versus cardial tumors (26 % vs. 18 %). Four- and 5-year survival for patients with adenocarcinoma and positive cervical nodes were 35 and 11 %, respectively.