In China, the predominant pathologic type of esophageal carcinoma is squamous cell carcinoma (ESCC), and many tumors locate in the middle thoracic. The overall 5-year survival rate of patients with lymph node metastasis (LNM) after surgical resection was only about 20% [1], and most of these patients died of tumor relapse. Recently, the therapeutic efficacy for ESCC had been greatly improved due to the establishment and promotion of multidisciplinary treatment model [2]. However, even for histological node-negative (pN0 stage) ESCC patients, a large number of these patients still experienced recurrence after operation [3]. So it was important to study the causes and characteristics of recurrence when designing therapeutic strategies, including operative procedures and combined therapies.

The purposes of this study were to explore the pattern and time of recurrence after extended radical esophagectomy with systemic lymphadenectomy, to identify the prognostic factors responsible for recurrence, and to discuss therapeutic strategies that might be helpful to decrease recurrence and improve survival.

112 patients were investigated and recurrence was recognized in 45 patients (40.2%) in the first 3 years after operation. The median disease-free interval until recurrence was 17.4 months (range: 3.5-26.8 months). Of these, 39 were men and 6 were women, and the median age was 56 years (range: 39–76 years).

The distribution of the sites of tumor recurrence

With respect to the locoregional tumor recurrence, 21 patients (15.3%) had recurrence in only cervical/supraclavicular lymph nodes, 13 patients (34.2%) had recurrence in only mediastinal lymph nodes, 6 (15.8%) in only abdominal lymph nodes. In cases of hematogenous recurrence, the main sites of recurrence were liver (6/7), lung (2/7), and bone (1/7). There was no one had recurrence in brain, pleura and omentum in our patients within 3 years after operation. Lymph node metastasis occured earlier than hematogenous recurrence (Table 2 and Figure 2).

Site of recurrence	No.	Percent (%)	Time of recurrence3(month)
Locoregional recurrence1	38
Cervical/supraclavicular node	21	55.3	14.9 (3.5 ~ 17.7)
Mediastinal node	13	34.2
upper	9	23.7	17.2 (4.8 ~ 23.2)
middle	3	7.9	19.2 (7.1 ~ 25.3)
lower	1	2.6	23.1
Abdominal node	6	15.8	18.4 (4.5 ~ 26.8)
Anastomotic	1	2.6	8.6
Hematogenous recurrence2	7
Liver	6	85.7	10.4 (5.2 ~ 20.4)
Lung	2	28.6	12.8 (9.4 ~ 16.2)
Bone	1	14.3	18.5
Others(brain/pleura/omentum)	0

¹Three patients experienced locoregional recurrence at more than one site.

²Two patients experienced hematogenous recurrence at more than one site.

³Mean (range).

The multivariate analysis for locoregional recurrence

The multivariate analysis by the Cox proportional hazards model demonstrated that Upper/middle thoracic location (p = 0.049, OR = 1.092) and pT3-4a stage (p = 0.017, OR = 3.296) were significant risk factors for locoregional recurrence of pN0 ESCC after operation (Table 3). We did not find a significant correlation between locoregional recurrence and gender, histological differentiation, or lymphadenectomy style.

Variable	B	SE	Wald	pValue	OR	95 % CI
Tumor location
(upper/middle versus lower)	1.087	0.557	3.808	0.049	1.092	1.088 ~ 2.383
Grade of differentiation
(G3 versus G1 + 2)	0.478	0.235	4.143	0.068	1.613	0.690 ~ 1.551
Depth of invasion
(pT3-4a versus pT1-2)	1.193	0.250	22.731	0.017	3.296	2.019 ~ 5.381

B: regression coefficient; SE: standard error; Wald: wald value; OR: odds ratio; CI: confidence interval.

The recurrence curve in pN0 patients with different pT stage was depicted in Figure 3 The log-rank test showed that there was a significant difference among the recurrence rates of these two groups (p = 0.000). The recurrence rate of the patients with upper/middle thoracic location was higher than the rate of lower thoracic location (p = 0.045) (Figure 4).

The 7th edition UICC TNM staging of esophageal cancer (2009) increased cell differentiation (G) and histological type (H) as a staging reference factor, also the pT3 stage (tumor invades adventitia) and pT4a stage (resectable tumor invading pleura, pericardium, or diaphragm) were distinguished [4]. With the wide application of the new version of the standard staging for esophageal cancer, the latest esophageal cancer treatment guidelines by National Comprehensive Cancer Network (NCCN) recommended for pN0 ESCC patients [5]: (1) pT1-4aN0M0 patients with ESCC, follow-up and observation after operation; (2) pT1-2N0M0 patients with adenocarcinoma, recommended follow-up and observation. However, some of the pT2N0M0 patients with high risk (poorly differentiated, lymphatic invasion, perineural invasion and <50 years of age), recommended postoperative chemoradiation; (3) pT3-4aN0M0 patients with adenocarcinoma were recommended postoperative chemoradiation or follow-up.

As we can see, prophylactic postoperative adjuvant therapy was not recommended for pT1-2N0M0 patients. However, there was no uniform international standards on whether pT3-4aN0M0 patients need postoperative adjuvant therapy or not, and what specific treatment should be chosen currently. It is worth noting that China, as a country with high incidence of esophageal cancer, nearly all of patients are squamous cell carcinoma. It is obviously different with the western countries in which the main histology of esophageal cancer was adenocarcinoma. Meanwhile, ESCC occur mostly in upper and middle thoracic, but adenocarcinoma occur mostly in lower thoracic. Whether from the perspective of anatomical or histological analysis, ESCC is more easy to have mediastinal lymph node metastasis even the neck and peritoneal metastasis than adenocarcinoma. Seen in this light, the foreign experience might not be suitable for China. Therefore, to study the risk factors for recurrence and metastasis in pN0 patients with ESCC, following suitable adjuvant treatment had important clinical significance for reducing recurrence and improving long-term survival.

Our cases showed that there was 40.2% of 112 pN0 ESCC patients had recurrence within 3 years after operation, with a median time (17.4 months) to recurrence, similar to the results reported in the literature [6]. Univariate analysis showed that recurrence closely correlated with tumor location, grade of differentiation, primary tumor stage (pT) and pathologic stage. The COX proportional hazards regression multivariate analysis confirmed that upper/middle thoracic location and pT3-4a stage were independent risk factors for postoperative locoregional recurrence. The pT stage (Wald = 22.731; OR = 3.296) was the most important risk factor on recurrence and metastasis of pN0 ESCC according to the Wald value. With increasing pT stage, the chance of recurrence and metastasis in pN0 esophageal cancer patients also increased significantly. The reason was that esophagus lacks serosa, once the tumor invades adventitia, not only the submucosal longitudinal lymphatic vessels but also lateral mediastinal lymphatic vessels originating from the muscle layer contribute greatly to the LNM [7]. In present study, the poorly differentiation (G3) did not have statistical significance with tumor recurrence might be due to small number of cases, but the p value (0.068) was close to 0.05. In addition, four patients with pathologically confirmed the presence of intralmural metastasis, three of them had cervical lymph node recurrence 2 years after operation and another appeared residual esophageal tumor recurrence 1 year after operation. It was suggesting that intralmural metastasis might be an important risk factor for recurrence of pN0 ESCC. Therefore, taking into account the possibility of multi-center origin in esophageal cancer and intralmural metastasis, we have to do the following in operation: (1) resection with sufficient length of esophagus; (2) complete resection of the lesion and the surrounding involved structures; (3) radical elimination of lymphatic tissue.

In the present study, the most common recurrence pattern of pN0 patients with ESCC after operation was locoregional recurrence (38/45, 84.4%) and LNM (37/38, 97.4%) was the main form, mainly in the cervicothoracic junction. Therefore, the cervicothoracic junction was the key position for preventing recurrence of pN0 patients with ESCC after operation. In addition, the difference of recurrence between two-field (thoracic and abdominal) lymph node dissections and three-field (cervical, thoracic and abdominal) lymph node dissections was not significant in our study. However, the literature had reported that three-field lymph node dissections could reduce recurrence and metastasis rate significantly [8]. The possible explanations were: Firstly, patients with pN0 ESCC might not need the three-field lymph node dissections; Secondly, there were statistical deviation due to small number of patients were given three-field lymph node dissections in this group, and this study was a retrospective analysis. Cervical B ultrasound was helpful to the diagnosis of LNM in cervicothoracic junction [9], combined with preoperative CT scan of chest/abdomen and positron emission tomography, establishing an accurate clinical staging was helpful to select the patients who do need three-field lymph node dissections [10]. We also took into account: (1) there were many important organs, large vessels in cervicothoracic junction and the region of recurrent laryngeal nerve was a relatively difficult position for systemic lymph node dissections because of higher requirements of surgical techniques; (2) three-field lymph node dissections had to face the bottleneck which can not further improve the long-term survival obviously. So, the postoperative radiotherapy and/or chemotherapy might be a good alternative.

Our data did not find independent risk factors for pN0 ESCC patients with hematogenous metastasis, suggesting that hematogenous metastasis was mostly influenced by the biological behavior of tumor cell rather than depth of tumor invasion, LNM and differentiation [11]. Thus, it was necessary to carry out prospective clinical study for evaluating whether vascular invasion, lymphatic invasion, and perineural invasion and so on were related with hematogenous metastasis or not.

In summary, we have confirmed that locoregional recurrence was the most common recurrence pattern of patients with pN0 ESCC within 3 years after operation. Upper/middle thoracic location and pT3-4a stage were independent risk factors for locoregional recurrence of pN0 ESCC after radical esophagectomy. For these high-risk groups, prophylactic radiotherapy for bilateral supraclavicular and upper mediastinal might be helpful to reduce the recurrence rate.