Clinical significance of esophageal invasion length for the prediction of mediastinal lymph node metastasis in Siewert type II adenocarcinoma: A retrospective single‐institution study

Abstract Aim This study investigated whether esophageal invasion length (EIL) of a tumor from the esophagogastric junction could be a possible indicator of mediastinal lymph node metastasis and survival in patients with Siewert type II adenocarcinoma. Methods One hundred and sixty‐eight patients with Siewert type II tumor who underwent surgery were enrolled. Metastatic stations and recurrent lymph node sites were classified into cervical, upper/middle/lower mediastinal, and abdominal zones. EIL was correlated with overall metastasis or recurrence in individual zones and with survival. Results Siewert type II patients with an EIL of more than 25 mm (>25 mm EIL group) had a higher incidence of overall metastasis or recurrence in the upper and middle mediastinal zones than those with an EIL of less than or equal to 25 mm (≤25 mm EIL group) (P = .001 and P < .001). Disease‐free and overall survival in the >25 mm EIL group were significantly lower than those of the ≤25 mm EIL group (P < .001). None of the Siewert type II patients with metastasis or recurrence in the upper and middle mediastinal zones survived for more than 5 years. Only an EIL of more than 25 mm was a significant preoperative predictor of overall metastasis or recurrence in the upper and middle mediastinal zones (odds ratio, 8.85; 95% CI, 2.31‐33.3; P = .001). Conclusion A multimodal‐therapeutic strategy should be investigated in Siewert type II patients once the tumor has invaded more than 25 mm to the esophageal wall.


| INTRODUCTION
The worldwide incidence of adenocarcinoma of the esophagogastric junction (EGJ) has been increasing in the past few decades. 1

-3 The
Siewert classification divides this entity into three subtypes according to its anatomical localization relative to the EGJ. 4 Siewert type I and type III tumors are usually treated as esophageal and gastric tumors, respectively. Because of the higher risk of mediastinal lymph node metastasis, a transthoracic and abdominal approach is frequently used for Siewert type I tumors, whereas an abdominal or transhiatal approach is commonly proposed for Siewert type III tumors. In contrast, Siewert type II tumors are considered "true" EGJ adenocarcinomas. 5 However, the optimal surgical approach for this tumor remains unclear. Although dissection of the lower mediastinal lymph nodes can be carried out by either a transthoracic or a transhiatal approach, only a transthoracic method can be used to dissect the upper and middle mediastinal lymph nodes. As an R0 resection is essential for long-term survival after surgery for EGJ adenocarcinoma, 6,7 the extent of lymph node dissection should be considered as well as the extent of esophageal or gastric resection in individual Siewert type II tumors.
Lymph node metastasis is known as a prognostic factor of esophageal cancer. Previous reports have indicated that a thorough lymph node dissection improves the survival of patients after esophagectomy. 8,9 However, the accuracy of preoperative assessments of positive lymph node metastasis in patients with EGJ adenocarcinoma using conventional diagnostic techniques, such as computed tomography (CT) and endoscopic ultrasonography (EUS), is not yet as high as needed. 10 Moreover, given the frequent discrepancy between clinical diagnosis and pathological findings, improvements in the preoperative assessment of positive lymph node metastasis, especially mediastinal lymph node metastasis, are key to optimizing the therapeutic strategy for Siewert type II tumors. Several pathological parameters have been shown to be risk factors for lymph node metastasis in patients with Siewert type II tumors. [11][12][13][14] However, because neoadjuvant chemotherapy or chemoradiotherapy has been introduced as a standard therapy for resectable EGJ adenocarcinoma, 15,16 a new clinical indicator that predicts mediastinal lymph node metastasis prior to treatment needs to be investigated for the prompt selection of surgical strategies, including the extent of lymph node dissection.
The surgical strategy for Siewert type II tumors is mainly determined by tumor invasion of the esophageal wall and the relative risk of upper and middle mediastinal lymph node metastasis. 17,18 Longitudinal lymphatic flow is very rich in the submucosal layer of the esophageal wall; 19,20 however, the incidence of mediastinal lymph node metastasis is very low when the tumor is predominantly located in the abdominal cavity and has not massively invaded the thoracic cavity. 21 23 The EGJ was defined as the distal end of the lower esophageal palisade vessels or as the upper ends of the gastric mucosal folds by endoscopic examination and was defined as the narrowest locus of the lower esophagus by esophagogastrography.

| Tumor classification
The EIL was subsequently determined as the distance from the EGJ to the proximal edge of the tumor. In some advanced circumferential tumors, the EIL was determined from the findings obtained by both Areas of lymph node metastases were recorded according to the lymph node stations adopted by the Japanese classifications. 24 Both the metastatic stations and the recurrent lymph node sites were classified into five lymph node zones: cervical, upper mediastinal, middle mediastinal, lower mediastinal, and abdominal. 25 The middle mediastinal zone and lower mediastinal zone were divided by the caudal margin of the inferior pulmonary vein. When either a metastasis or a recurrence was detected in any lymph node of each zone, that lymph node zone was considered to be positive for tumor involvement. Three different approaches were adopted based on the results of preoperative tumor examinations: a right-sided transthoracic and abdominal approach (RTT), a left-sided thoracoabdominal approach (LTA), and a transhiatal via abdominal approach (TH). Generally, RTT was selected for advanced tumors or tumors with clinical lymph node metastasis. LTA and TH were selected for mainly superficial tumors, and left thoracic incision was carried out to secure the proximal surgical margin. Extent of lymph node dissection basically depended on the surgical approach: three-or two-field lymph node dissection was used for an RTT, lower to middle mediastinal and abdominal lymph node dissection was used for an LTA, and abdominal and partial lower mediastinal lymph node dissection was used for a TH. The gastric tube was preferably used for reconstruction when an RTT was carried out, and a jejunal interposition was selected when an LTA or a TH was carried out. Respiratory complications included pneumonia, atelectasis, and empyema. The clinical course after surgery was followed every 3 months for the first year, and then every 6 months up to 5 years. In addition to the physical examination and serum tumor markers, CT scan was carried out every 6 months, and endoscopy was carried out once per year to survey tumor recurrence.

| Statistical analysis
Chi-squared test and Fisher's exact test were used to assess the categorical variables, and a t test and a Mann-Whitney U test were used to assess continuous variables. Survival curves were generated using the Kaplan-Meier method, and a log-rank test was used to compare survival. Receiver operating characteristic (ROC) curves were generated to decide the cut-off value for the EIL in Siewert type II tumors to predict the risk of mediastinal lymph node metastasis or recurrence. Using the frequency of overall metastasis or recurrence and the 5-year overall survival (OS) in each zone, an efficacy index was established to evaluate the benefits of mediastinal lymph node dissection in patients with Siewert type II tumor. 25 A multivariate logistic regression analysis was carried out for the selection of predictive clinical variables that affected upper and middle mediastinal lymph node metastasis or recurrence. The model did not include variables in which the P-value was not significant in a univariate analysis and in which multicollinearity with the EIL was recognized. The analysis was carried out using SPSS software (version 22; SAS Institute, Cary, NC, USA), and the tests were two-sided with a significance level <.05.

| Characteristics of patients with Siewert type II adenocarcinoma
Characteristics of patients with Siewert type II tumors are summarized in Table 1. There were 63, 21, and 84 patients categorized as having clinical T1, T2, and T3 tumors, respectively. Also, there were 102, 41, 24, and one patients categorized as having clinical N0, N1, N2, and N3 lymph node metastasis, respectively. One patient had supraclavicular lymph node metastasis. Based on the diagnosis of these categories, patients were clinically classified as belonging to stage IA (n = 61), IB (n = 15), IIA (n = 25), IIB (n = 7), IIIA (n = 35), IIIB (n = 23), IIIC (n = 1), and IV (n = 1). Pathological lymph node metastasis was found in 88 (52.4%) patients with Siewert type II tumors. Rate of pathological lymph node metastasis was higher than the rate of clinical lymph node metastasis. Incidence of any lymph node recurrence was 8.9% in Siewert type II tumors. However, as a result of the different rate of mediastinal lymph node dissection, immediate comparison between pathological metastasis or postoperative recurrence of the lymph node and other clinicopathological parameters might give misleading information. Therefore, we combined the metastatic stations and recurrent lymph node sites and classified them into five lymph node zones (Table S1) Table 3). Incidence of overall metastasis or recurrence was increased with the EIL of Siewert type II tumors, and this tendency was especially apparent in the mediastinal zones. Among patients with an EIL of <20 mm (n = 86), upper or middle mediastinal lymph node metastasis or recurrence occurred in one patient only (1%). In contrast, in patients with an EIL of more than 30 mm (n = 50), upper and middle mediastinal lymph node metastasis or recurrence occurred in 10 patients (20%). ROC curve for EIL in patients with Siewert type II tumor was generated to predict the rates of upper and middle mediastinal lymph node metastasis or recurrence, and the cut-off EIL value was determined to be 25 mm with an area under the curve of 0.83 according to the ROC curve (sensitivity, 80.8%; specificity, 72.3%) (Figure 1). There were 118 patients whose EIL was less than or equal to 25 mm (≤25 mm EIL group) and 50 patients whose EIL was more than 25 mm (>25 mm EIL group).
Patient characteristics were compared between the ≤25 mm EIL group and the >25 mm EIL group (Table 4). Tumor stages in the >25 mm EIL group were significantly more advanced than those in the KOYANAGI ET AL.
| 189 ≤25 mm EIL group (P < .001). Lymphatic and venous invasions were significantly more frequent in the >25 mm EIL group (P < .001 and P = .002, respectively). Overall metastasis or recurrence was compared between the ≤25 mm EIL group and the >25 mm EIL group ( Table 5). Rates of overall metastasis or recurrence in the cervical, upper, middle, lower mediastinal, and abdominal zones were significantly higher in the >25 mm EIL group than in the ≤25 mm EIL group (P = .007, P = .01, P < .001, P = .002, and P < .001, respectively).

| Overall survival and risk assessment of mediastinal lymph node metastasis in Siewert type II tumors
Median duration of the follow-up period for all the patients was 60 months. Estimated 5-year disease-free survival (DFS) rates for the ≤25 mm EIL group and the >25 mm EIL group were 67.1% and 41.3%, respectively (P < .001) (Figure 2A). Estimated 5-year OS rates for the ≤25 mm EIL group and the >25 mm EIL group were 66.8% and 40.9%, respectively (P < .001) ( Figure 2B). The 5-year OS rates of the Siewert type II patients without overall metastasis or recurrence in the upper mediastinal, middle mediastinal, lower mediastinal, and abdominal zones were 64.7%, 62.8%, 64.5%, and 52.3%, respectively ( Figure 3). In contrast, the rates of those with overall metastasis or recurrence in the upper mediastinal, middle mediastinal, lower mediastinal, and abdominal zones were 0%, 0%, 37.5%, and 41.9%, respectively, and these differences between patients with and those without overall metastasis or recurrence in each of the zones were significant for Siewert type II tumors (P < .001, P < .001, P = .013, and P < .001, respectively). Patients with Siewert type II tumors who had overall metastasis or recurrence in the cervical, upper, and middle mediastinal zones did not survive for more than 5 years after surgery. Efficacy indexes of the lower mediastinal lymph node and the abdominal lymph node were 5. 6