Introduction
Although the incidence trend decreased monotonically over the past decades, gastric cancer still accounts for terrible cancer-related deaths and remains an enormous health burden globally [
1‐
3]. In recent decades, more and more population-based researches that had illustrated the incidence of the adenocarcinoma in esophagogastric junction (AEG) presented a distinctive increased trend [
4‐
6]. A retrospective study from our institution had also indicated the proportion of AEG among surgical patients had increased from 22.3 to 35.7% during the past 25 years [
7]. When contrasted with the distal gastric adenocarcinoma, AEG might have its own heterogeneity and issued in a relatively worse prognosis [
8]. Due to the distinguishing anatomic location, the classification of AEG had been always in debate during the past decades. As we all know, a rigorous and coherent staging system for AEG was crucial for clinicians to pick the optimal subsequent treatment regime [
9,
10]. An accurate classification system AEG could also help clinicians make more suitable follow-up tactics and predict potential prognosis for these patients [
11].
Currently, the American Joint Committee on Cancer (AJCC) Tumor-Node-Metastasis (TNM) staging system was the most common system that applied to solid tumors including gastric and esophageal cancers [
12]. Numerous editions of staging system had been successively updated to reflect the prognosis as veritably as possible [
13]. In the Fall of 2016, AJCC released the 8th edition of TNM classification which included many important modifications especially the redefinition for AEG: cancers with EGJ invasion that have their epicenter within the proximal 2 cm of the EGJ (Siewert type I/II) are to be staged as TNM-EC. Cancers whose epicenter is more than 2 cm distal from the EGJ, even if the EGJ is involved, would be staged by TNM-GC [
12]. Thus, it could be seen that AEG II was still ascribed to TNM-EC.
The past seven editions of AJCC staging system ascribed AEG to TNM-EC system [
14]; however, some researches had indicated that the staging scheme of AEG should be according to TNM-GC [
15,
16]. Furthermore, some other reports had manifested that neither of these two systems could factually reflect prognosis of AEG and a new staging system should be introduced to this entity [
17,
18]. In addition, the database from China for the new edition was relatively rare, whether the 8th staging system could really represent the prognosis of Chinese patients was still in suspicion [
19,
20]. Even a further definition for the classification of AEG was established in the 8th edition, these two schemes which could better describe the prognosis of AEG especially for Chinese population was still ambiguous. To sum up, the classification system for AEG was always in controversy, gastric or esophageal scheme, which was more suitable for the classification of AEG (especially for type II) remained in discussion.
To the best of our knowledge, there were no large-sample reports to compare the classification of AEG according to the 8th edition of TNM-GC and TNM-EC. In consideration of these issues, we established a predictive nomogram by retrospectively analyzing database from two high-volume institutions in China to compare which system could better describe the classification of AEG (especially for type II) between TNM-GC and TNM-EC.
Discussion
Recently, there were many reports that concern about the increased trend of AEG and indicated that its unique features differ from common gastric and esophageal adenocarcinomas [
4‐
6,
27]. A veracious staging system for AEG was the primary basis for evaluating prognosis, making suitable strategy on stage-specific treatments [
17]. However, there were also many different opinions on the classification of AEG that still left the propensity of the staging system of AEG in ambiguity [
15‐
17,
28]. The 8th edition made the definition of AEG more clear and still ascribed AEG II to TNM-EC [
12,
19,
29]. However, whether the definition was indeed adequate for AEG II especially for Chinese population was still unknown. Our study demonstrated that TNM-GC had a significant advantage over TNM-EC with regard to the correlation between HR and stage, the separation of the survival among different stages in both training and validation cohorts. The effectiveness and accuracy of TNM-GC were better than TNM-EC; the stratified analysis also indicated the consistent results for Siewert type II in both cohorts. These results might have an influence on subsequent TNM staging system revision for AEG.
A monodirectional increased trend of HRs was observed in TNM-GC in both training and validation cohorts; however, when classified by TNM-EC, only the fluctuating increase was found between stages IIB and IIIB. These might indicate that TNM-EC was insufficient in describing the survival trend in these stages. One possible explanation was that patients in stage IA and IB were too rare to produce a credible value of HRs in these three subgroups when in terms of TNM-EC. On the other hand, the survival has no significance from stage IA to II A in TNM-EC in both cohorts. The similar outcome was also observed from a study in Japan [
15]. An adverse survival curve was detected in stages IC, IIA, IIB and IIIA in both cohorts when according to TNM-EC. The stratification analysis for AEG II also indicated similar outcomes. All these above results might demonstrate that the discrimination and factuality of TNM-EC were insufficient for AEG type II. The following comparison between two Cox multivariable models further ascertains that TNM-GC plays a more important role in predicting the prognosis for AEG type II in both cohorts. The prognostic model which included TNM-GC might more efficient than that included TNM-EC, a smaller AIC and larger C-index were dug out in both training and validation cohorts and consistent outcomes were detected in AEG II in both cohorts.
As many previous reports, the different surgical approaches combined with different lymph node dissection could bring different survival outcomes. Resection of more lymph nodes could bring to a better prognosis and a more accurate survival estimation [
30,
31]. Recent researches further indicated that the extent of lymphadenectomy is intimately related to the overall survival of patients with AEG [
32]. In our study, when compared with transthoracic approach, a deficient upper mediastinal lymphadenectomy may occur via transhiatal approach. The scanty number of harvest lymph nodes might also cause stage migration. On the other side, the lymph node metastatic manner and tumorigenesis in this region may be different from genuine esophageal adenocarcinoma [
33]. These may partly paraphrase the discrimination of survival outcomes from esophageal adenocarcinoma. For AEG II patients, the issue of surgical approach (transthoracic or transhiatal) and lymphadenectomy had been always in debate [
31,
34,
35]. Different surgical approaches were performed in different departments and combined with various lymphadenectomies. These could result in different survival outcomes for AEG II patients [
31,
34]. Considering the potential bias brought by surgical approaches, we exclude AEG II patients who underwent transthoracic resection. In our study, the outcomes could indicate that TNM-GC was superior over TNM-EC for AEG II patients who underwent transhiatal approach with sufficient abdominal lymph node dissection.
In line with some former reports [
16,
36], the classification of AEG type II should be considered as a part of gastric cancer. Our study also illustrated that TNM-GC manifested a superior heterogeneity compared with TNM-EC. The anatomical and adjoining structures of type II were more semblable to stomach instead of esophagus and the structure of gastric wall in the position of type II was same with stomach [
37]. On the other hand, the location of type II that squamous epithelium transformed to glandular epithelium is rather different from the squamous epithelium of esophagus. Different epithelial ingredients with different tumorigenesis might induce discrepant prognosis. However, the pathological components in type II have become glandular epithelium which was more similar to typical gastric epithelium [
37]. The tumorigenicity in this position might have something in common with typical gastric adenocarcinoma. A recent basic research also indicated that the esophageal adenocarcinomas had strong similarity with the chromosomally unstable variant of gastric adenocarcinoma, suggesting that these cancers including AEG were more similar with gastric adenocarcinoma in molecular level or should be considered as a single disease entity [
38]. In summary, Siewert type II might have more common with gastric adenocarcinoma from the point of anatomy and pathology.
The former research had found more proportion of undifferentiated type in gastric cardia cancer leads to worse prognosis [
39]. The histological grade was required for the final stage in the TNM-EC in the 8th edition. However, Kim et al. reported that histological factor might recede the prediction of patient survival for AEG [
40]. The groups with G1/2 differentiated adenocarcinoma indicated better survival than those with G3 in training cohort. Our report indicated that histological grade had a significant correlation with prognosis in univariate analysis in both training (HR = 1.359) and validation (HR = 1.410) cohorts; however, histological grade was not an independent prognostic factor for AEG after multivariable analysis in both cohorts. Consequently, whether histologic grade should be applied to the final staging system of AEG is required to be prudently considered further.
In summary, many factors may have different impacts on the prognosis of AEG, the existent 8th TNM staging system might still have some insufficiency in predicting the prognosis. Therefore, a veracious staging system that is suitable for AEG accurately needs to have further consideration [
32]. As mentioned in our research, both schemes were defective in representing the survival of stage IA–IIA. One reason was the relatively low proportion of early staging cancer in China and the number of patients in these stages was too rare to reflect the distinctiveness. An initiate staging system specific for AEG was also demanded to describe the prognosis exactly in the future.
This research has several limitations. In the first place, there was no Siewert type I cancer enrolled. In the second, we did not include AEG II patients who underwent transthoracic approach, all the patients in our study were only treated by transhiatal gastrectomy and not received upper mediastinum lymphadenectomy. Third, the number of patients in stages I and II was relatively small to reflect the discriminative prognosis of AEG and the stage relatively concentrated to stage III in our study which may also have an impact on the overall survival outcomes.
Acknowledgements
The authors thank the substantial work of Volunteer Team of Gastric Cancer Surgery (VOLTGA) West China Hospital, Sichuan University, China, for the establishment of gastric cancer database and the continual follow-up. The authors thank Prof. Yong Yuan and Long-Qi Chen from Department of Thoracic Surgery, West China Hospital, Sichuan University, for providing data for the additional analysis in the process of revision.