Abstract
Aim: To define the optimal extent of resection for esophagogastric junction (EGJ) carcinoma. Patients and Methods: We retrospectively reviewed medical records of 193 patients with EGJ adenocarcinoma or squamous cell carcinoma who underwent surgery at the Kitasato University. An index was calculated to evaluate the therapeutic value of lymphadenectomy. Results: The following factors were identified as independent predictors of poor survival: (y)pT3-4, (y)pN3, ly2-3, no performance of splenectomy and R1-2. Although metastases were found in mediastinal lymph-nodes in patients with esophageal invasion of ≤30 mm, the index was 0 for all mediastinal lymph-nodes. By contrast, in patients with esophageal invasion of >30 mm, the index was 13.9 for the No. 110 nodes, which was the second highest after the index for the No. 1 nodes. Conclusion: In EGJ cancer patients with esophageal invasion of >30 mm, aggressive lower mediastinal lymphadenectomy with R0 resection is required to obtain the best result.
The incidence of adenocarcinoma of the esophagogastric junction (AEG) is rising in Western countries (1, 2) and Japan (3, 4). However, the prevalence of squamous cell carcinoma (SCC) remains higher in Japan than in Western countries.
Carcinoma of the esophagogastric junction (EGJ) is known to carry an early risk of extensive metastases to the thoracic and abdominal lymph nodes. Lymph node dissection in these regions is considered important but is very invasive. The optimal extent of lymph node dissection, thus, remains controversial.
Siewert et al. recently proposed a classification system for AEG and discussed the characteristics and treatment of AEG according to disease type. They proposed that type I AEG should be treated similarly to distal esophageal cancer, whereas patients with type II and III AEG should undergo transhiatal total gastrectomy, lower esophagectomy, lower mediastinal lymphadenectomy and extended (D2) lymph node dissection comparable to that performed in patients with gastric cancer (5). Conversely, in a study in which more than half of the patients had Siewert type II AEG, esophagectomy was associated with better outcomes than gastrectomy (6). In Japan, Sasako et al. reported that a left thoracoabdominal approach does not improve survival after an abdominal–transhiatal approach and leads to increased morbidity in patients with adenocarcinoma of the gastric body or cardia with esophageal invasion of ≤30 mm (7). However, these studies included only patients with adenocarcinoma; they did not include SCC, which continues to have a high incidence in the Eastern countries.
We, herein, studied patients with EGJ carcinoma, including SCC. Lymph node metastasis, nodal recurrence patterns and prognostic factors were analyzed according to Siewert's classification. We also evaluated the therapeutic value of lymph node dissection to define the optimal extent of lymphadenectomy.
Patients and Methods
Definition of EGJ carcinoma. EGJ carcinoma was defined as carcinoma that invaded the EGJ with a tumor epicenter located ≤5 cm from the EGJ. Tumors were classified according to Siewert's classification as follows: type I, tumor epicenter located 1 to 5 cm above the EGJ; type II, tumor epicenter located 1 cm above to 2 cm below the EGJ; and type III, tumor epicenter located 2 to 5 cm below the EGJ. The location of the epicenter and the proximal extent of the tumor were comprehensively evaluated based on findings obtained by an upper gastrointestinal series and upper gastrointestinal endoscopy. In patients who received preoperative chemotherapy, tumors were classified based on the characteristics before chemotherapy.
Patients and clinicopathological evaluation. From January 1997 through December 2012, a total of 3,004 patients underwent surgery for esophageal cancer or gastric cancer at Kitasato University. Of these patients, 195 (6.5%) underwent surgical resection of EGJ carcinoma. Two of these patients had neuroendocrine carcinoma and were excluded. The remaining 193 patients (6.4%, 193/3,004) were studied. The median follow-up period was 36 months (inter quartile range, 18-61 months).
Tumor stage was classified according to the International Union against Cancer TNM staging system, 7th edition (8). Lymph node stations were classified according to the Japanese Classification of Gastric Carcinoma (9) and the Japanese Classification of Esophageal Carcinoma (10). The cervical paraesophageal nodes (No. 101) and supraclavicular nodes (No. 104) were classified as cervical lymph nodes; the upper thoracic paraesophageal nodes (No. 105) and thoracic paratracheal nodes (No. 106) were classified as upper mediastinal lymph nodes; the subcarinal nodes (No. 107), middle thoracic paraesophageal nodes (No. 108) and main bronchus nodes (No. 109) were classified as middle esophageal lymph nodes; and the lower thoracic paraesophageal nodes (No. 110), supradiaphragmatic nodes (No. 111) and posterior mediastinal nodes (No. 112) were classified as lower thoracic lymph nodes.
Adenocarcinoma was histologically classified according to the definitions of the Japanese Classification of Gastric Carcinoma. Well- and moderately-differentiated tubular adenocarcinomas and papillary adenocarcinomas were classified as differentiated-type adenocarcinoma, whereas poorly-differentiated adenocarcinomas, signet-ring cell carcinomas and mucinous carcinomas were classified as undifferentiated-type adenocarcinoma.
Perioperative transfusion was defined as allogeneic blood transfusion performed during the operation or within the first 2 postoperative days, as previously reported (11).
The patients were divided into two groups: those with esophageal invasion of >30 mm and those with esophageal invasion of ≤30 mm. This classification was based on the results of the JCOG9502 study (7).
The present study was conducted in accordance with the Declaration of Helsinki and approved by the Research Ethics Committee of Kitasato University School of Medicine. The requirement for informed consent was waived because of the retrospective study design.
Surgical procedures. Surgical procedures were determined on the basis of tumor location and length of esophageal invasion. A right thoracic approach (RTA) was used to perform subtotal esophagectomy and mediastinal lymphadenectomy through a right thoracotomy, while gastric tube reconstruction was performed through a laparotomy. A left thoracic approach (LTA) through a left thoracotomy and laparotomy and a transhiatal approach (THA) after wide splitting of the esophageal hiatus were used to perform total gastrectomy, distal esophagectomy, D2 lymphadenectomy and lower mediastinal lymphadenectomy. When the tumor invaded the greater curvature of the stomach, splenectomy was performed to dissect splenic hilar lymph nodes completely. For cT1 carcinoma, proximal gastrectomy and jejunal interposition or esophagogastric anastomosis were performed. Because video-assisted thoracoscopic esophagectomy was performed through an RTA in only five patients, these patients were included in the RTA group for analysis.
Chemotherapy. Preoperative chemotherapy was given to 35 patients, 17 with resectable disease and 18 with unresectable disease. The most commonly used regimen was a combination of docetaxel, cisplatin and S-1 (DCS). The treatment schedule for DCS has been previously described (12).
Statistical analysis. Overall survival (OS) was calculated from the date of surgery or from the date of starting chemotherapy in patients who received preoperative chemotherapy. The Student's t-test and Fisher's exact test were used to analyze continuous and categorical variables, respectively. Survival was calculated by the Kaplan–Meier method. Univariate analyses of prognostic factors for OS were performed using log-rank tests. Factors with p<0.10 on univariate analysis were subjected to multivariate analysis using the Cox's proportional-hazards model to identify independent prognostic factors. All calculations were performed using the JMP® 10 software (SAS Institute Inc., Cary, NC, USA) and p-values of <0.05 were considered to indicate statistical significance.
Therapeutic value of lymph node dissection. To evaluate the therapeutic value of lymph node dissection, we used a previously described method (13) in patients with R0 (no) residual tumors. In brief, the frequency of metastasis to each station was determined by dividing the number of patients with metastasis at that station by the number of patients in whom the station was dissected. The 5-year OS rate of patients with metastasis at each nodal station was calculated irrespective of the presence or absence of metastases at other nodal stations. An index of the benefit gained by dissection of each station was calculated by multiplying the frequency of metastasis at the station by the 5-year survival rate of patients with metastasis at that station.
The index was also evaluated separately in patients with esophageal invasion of >30 mm and in patients with esophageal invasion of ≤30 mm.
Results
Patients' characteristics and stage distribution. The patients' characteristics are listed in Table I. The median number of dissected lymph nodes was 42, 40 and 32 in the RTA group, the LTA group and the THA group, respectively. The number of dissected lymph nodes was significantly less in the THA group than in the RTA group (p=0.039). The pathological stage distribution is shown according to the tumor types in Figure 1a-c. Most cases of early cancer were type I or II tumors, while early disease was rare among type III tumors. Type III tumors tended to be carcinomas associated with incurable factors, such as stage IV disease.
Surgical approach, morbidity and mortality. Surgical approaches are shown according to tumor types in Figure 1d-f. Surgery was performed via an RTA in nearly all patients with type I tumors. A THA was used in many patients with type II or III tumors.
Surgical complications classified by the surgical approach and Clavien–Dindo classification (14, 15) are shown in Table II. Grade III surgical complications were more frequently found in patients who underwent an RTA than in those who underwent a THA (47.5 vs. 17.6%, respectively; p<0.001). Operative mortality was found in one patient who underwent an RTA but in no patients who underwent an LTA or THA.
Patients' characteristics.
Recurrence and metastasis to the cervical and mediastinal lymph nodes. The frequencies of recurrence and metastasis to the cervical and mediastinal lymph nodes are shown in Figure 2. These frequencies were based on the numbers of patients with metastases in dissected lymph nodes and/or the numbers of patients with nodal recurrence during follow-up. Compared to type II and III tumors, type I tumors were associated with higher rates of recurrence and metastasis in the mediastinal lymph nodes, particularly the lower mediastinal lymph nodes. Both SCC and adenocarcinoma were associated with recurrence and metastasis to the mediastinal lymph nodes but the frequency was significantly higher for SCC (p=0.010).
Five (18%) out of 28 patients with T1b disease and 12 (60%) of 20 patients with T2 disease had recurrent and/or metastatic lymph nodes. Among these, one patient (4%) with T1b disease and one patient (5%) with T2 disease had lower mediastinal metastatic lymph nodes. These two patients had tumors with esophageal invasion of >30 mm.
Survival analysis according to histology and the tumor types. Type II tumors were associated with a trend towards better OS than type I or III tumors (p=0.19) (Figure 3b). Among type I and II tumors, survival was similar between patients with SCC and those with adenocarcinoma (Figure 3c, d). Therefore, we did not evaluate SCC and adenocarcinoma separately in this study.
Stage distribution (a, b, c) and surgical approach (d, e, f) stratified according to tumor type.
Frequency of recurrence and metastasis to the cervical and mediastinal lymph nodes (LN) in all cases (a), adenocarcinoma cases (b) and squamous cell carcinoma cases (c).
Overall survival according to tumor type (a, b) and histological type in type I cases (c) and type II cases (d).
Prognostic factors identified in univariate and multivariate prognostic analyses. The results of univariate and multivariate analyses of prognostic factors for OS are shown in Table III. The following factors were identified as independent predictors of poor survival: a pathological tumor stage of (y)pT3-4, a pathological nodal stage of (y)pN3, lymphatic invasion of ly2-3, no performance of splenectomy and residual tumor of R1-2.
Sub-analysis of prognostic factors in type II and III EGJ carcinoma. A joint analysis of type II and III tumors was performed because the surgical approaches and the status of nodal metastasis and recurrence were similar between patients with type II and III tumors (Figures 1, 2). The results of analysis of prognostic factors for OS are shown in Table IV. The following factors were found to be independent predictors of poor survival: a pathological nodal status of (y)pN3, vascular invasion of v0-1, performance of perioperative transfusion, performance of thoracotomy and residual tumor of R1-2.
Recurrence pattern of the patients who underwent thoracotomy and those who did not undergo thoracotomy. Out of the 58 patients who underwent thoracotomy, 26 had hematogenous metastasis including 11 in the liver, 7 in the lung, 5 in the bone and 3 in the brain. Out of the 135 patients who did not undergo thoracotomy, 30 had hematogenous metastasis including 15 in the liver, 6 in the lung, 6 in the bone, 2 in the adrenal grand and 1 in the brain. The incidence of hematogenous metastasis in patients who underwent thoracotomy was significantly higher than that in those who did not undergo thoracotomy (p=0.002).
Morbidity classified by surgical approach and Clavien–Dindo classification.
Estimated benefit from lymphadenectomy. Table V shows the proportion of metastases according to lymph node station and the corresponding index of the estimated benefit from lymph node dissection in 155 patients who underwent R0 surgery.
In patients with esophageal invasion of ≤30 mm, metastases were found in mediastinal lymph nodes, such as Nos. 110 and 112, but the index of the estimated benefit from lymph node dissection was 0 for all mediastinal lymph nodes. Metastases to the mediastinal lymph nodes were also found in patients with esophageal invasion of >30 mm. The index was 13.9 for the No. 110 nodes, which was the second highest after the index for the No. 1 nodes. The index for the No. 110 nodes in patients with type I, II and III tumors was 12.9, 5.3 and 0, respectively. Two patients had metastasis to the No. 110 nodes and survived for >5 years: one had type I SCC and underwent surgery by an RTA and the other had type II adenocarcinoma and underwent surgery by an LTA.
Discussion
Similar to the results of previous studies (16), the benefit of lymph node dissection was high in the No. 1, 3 and 7 nodes; a benefit was also obtained in the No. 110 nodes. In particular, in patients with esophageal invasion of >30 mm, the index of the estimated benefit of lymph node dissection of the No. 110 nodes was 13.9, which was the second highest value after the No. 1 nodes. In patients with esophageal invasion of ≤30 mm, however, the benefit index was 0 for the No. 110 nodes; all patients with lymph node metastases in the middle and upper mediastinal regions died of their disease within 5 years.
In the literature, the risk for lymph node metastasis of esophageal cancer in sm1 disease ranges from 8.3 to 21%, that in sm2 disease ranges from 20 to 36% and that in sm3 disease ranges from 19 to 50% (17-19). Thus, aggressive nodal staging and resection is quite important in this cohort. In this study of EGJ carcinoma, however, only 4 and 5% of patients with T1b and T2 disease, respectively, had metastatic lower mediastinal lymph nodes. By contrast, when we restricted patients to those with T1b and T2 tumors that invaded the esophagus to >30 mm, the rate of metastasis to the lower mediastinal lymph nodes was as high as 14% (1/7) and 33% (1/3), respectively. The JCOG9502 study did not demonstrate superiority of the LTA over the THA in patients with esophageal invasion of ≤30 mm (7); our results support this finding. In patients with esophageal invasion of ≤30 mm, aggressive lower mediastinal dissection was minimally beneficial. The achievement of negative resection margins (R0 surgery) and reduced postoperative complications appear to be more important. In patients with esophageal invasion of >30 mm, however, aggressive lower mediastinal dissection should be performed if R0 surgery is feasible.
Recent publications have documented the importance of lymphadenectomy for esophageal cancer (20-22). Altoki et al. reported that the minimum required number of dissected nodes for staging was 17 and that Ivor Lewis esophagectomy has a potential survival benefit over transhiatal esophagectomy (23). In our current study, the median number of dissected lymph nodes was 32, 40 and 42 in the THA, LTA and RTA, respectively. The number of dissected lymph nodes was significantly less in the THA than in the RTA. However, the number of dissected nodes in the THA is approximately twice the minimum required number. If we consider the invasiveness and benefit of mediastinal lymph node dissection by thoracotomy, we cannot confirm the superiority of the RTA for all types of EGJ cancer.
Prognostic analysis in patients with esophagogastric junction (EGJ) cancer.
Thoracotomy was a poor prognostic factor in patients with type II and III cancer. In terms of recurrence patterns, the incidence of hematogenous metastasis in patients who underwent thoracotomy was significantly higher than that in patients who did not undergo thoracotomy. The excessive surgical stress associated with thoracolaparotomy has been shown to remarkably enhance tumor metastasis by a mechanism involving immunosuppression in an experimental model (24). The results of our study may have also been related to some type of immunosuppression. Moreover, the patients who underwent surgery by an LTA or RTA tended to develop severe surgical complications. For this reason, an LTA or RTA should be avoided if R0 resection can be achieved by a THA.
In terms of minimally-invasive techniques in esophageal cancer surgery, large nationwide studies in Japan (25) and UK (26) did not show clear benefits. They reported that morbidity rate was significantly higher in the minimally invasive esophagectomy (MIE) group than in the open esophagectomy group. On the other hand, meta-analyses (27, 28), retrospective reviews (29, 30) and a randomized multi-center trial (31), which were conducted in high-volume centers, showed that MIE was associated with lower morbidity and decreased pulmonary complication than open esophagectomy. These findings show that MIE could be safely performed not in every institution but in selected high-volume centers. Therefore, the thoracic approach might have a better prognosis than the current results, if minimally-invasive approaches were more frequently used in selected high-volume Centers.
Prognostic analysis of EGJ type II and III cancer.
Our study of recurrence and lymph node metastasis in patients with EGJ cancer has shown that SCC is associated with a significantly higher rate of mediastinal lymph node metastasis and/or recurrence than adenocarcinoma. This trend was shown to be particularly strong in the lower mediastinum. It was also reported that SCC is associated with a higher risk of mediastinal lymph node metastasis than adenocarcinoma in patients with EGJ carcinoma (32), consistent with our results. In our study, however, more patients with adenocarcinoma had early cancer than SCC. Therefore, our current data do not allow us to conclude that SCC is more malignant and associated with a higher risk of mediastinal metastasis than adenocarcinoma.
Nevertheless, SCC could have a high risk of metastasis to the mediastinal lymph nodes, and many type I tumors were SCC; thus, right thoracic subtotal esophagectomy with gastric tube reconstruction was considered an appropriate procedure for patients with type I tumors.
Because the dissection index of the No. 4d, No. 5 and No. 6 nodes were 0, we believe that total gastrectomy is not necessarily needed for EGJ cancer patients. In patients with esophageal invasion of ≤30 mm, thoracotomy should only be performed if a tumor-free esophageal margin cannot be secured. Given the potential effect of thoracotomy on immunosuppression, securing an esophageal resection margin by video-assisted thoracoscopic surgery (VATS) might be considered for patients in whom a negative esophageal resection margin cannot be obtained.
For locally advanced EGJ cancer, multimodal therapy has been performed. Magic trial demonstrated that patients who received perioperative chemotherapy with epirubicin, cisplatin and 5-fluorouracil (5FU) had significantly better OS as compared to those who received surgery alone (33). A French Trial demonstrated that patients who received perioperative chemotherapy with 5FU and cisplatin had significantly better OS as compared to those who received surgery alone (34). These two trials were targeted for patients with adenocarcinoma and included 26% and 64% of patients with EGJ cancer, respectively. More recently, the Cross trial demonstrated significant improvements of OS in patients who received chemoradiotherapy with carboplatin and paclitaxel, plus 41.4Gy as compared to those who received surgery alone (35). In this trial, 23% of all the patients had SCC and 24% of all the patients had EGJ cancer. Therefore, for patients with locally advanced EGJ cancer, neoadjuvant chemoradiotherapy is the treatment of choice in particular for SCC. However, adjusted hazard ratios demonstrated no improvement in survival for patients with node-positive disease. In addition, although chemoradiotherapy reduced the risk of hematogenous metastases, distant metastasis remained the most common form of disease recurrence (36). Improving local control may not lead to a survival advantage for patients with EGJ cancer.
Metastatic rate and index of estimated benefit from lymph node dissection.
Our study had several limitations. The analysis was based on retrospective data collection at a single Center with a small sample size. Treatment policies were not necessarily standardized and the demographic characteristics of the patients varied considerably. In particular, an LTA was rarely used after the results of the JCOG9502 study were reported. The regimens and indications for preoperative chemotherapy were also not standardized. Moreover, neoadjuvant chemoradiotherapy, which is thought to be the standard therapy for EGJ cancer, was not given to any patients.
In conclusion, type I tumors, particularly SCC, are frequently associated with mediastinal lymph node metastasis. Surgery by an RTA, similar to that for esophageal cancer, is, thus, considered appropriate. In patients with type II or III tumors, particularly those with esophageal invasion of ≤30 mm, extended lymphadenectomy by a THA should be performed. In patients with esophageal invasion of >30 mm, R0 resection and aggressive lower mediastinal lymphadenectomy may be required to obtain the best result. Thoracotomy should be avoided if R0 resection can be achieved by a THA. Further multicenter, randomized controlled trials of patients with EGJ cancer with esophageal invasion of >30 mm should be performed to confirm our results.
Footnotes
-
Conflicts of Interest
None declared.
- Received September 13, 2014.
- Revision received October 15, 2014.
- Accepted October 22, 2014.
- Copyright© 2015 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
