- © 2001 by American Society of Clinical Oncology
Esophageal Cancer: The Mode of Lymphatic Tumor Cell Spread and Its Prognostic Significance
- Stefan B. Hosch,
- Nikolas H. Stoecklein,
- U. Pichlmeier,
- Alexander Rehders,
- Peter Scheunemann,
- Axel Niendorf,
- Wolfram T. Knoefel and
- Jakob R. Izbicki
- From the Department of Surgery, University of Hamburg; the Institute of Mathematics and Computer Science in Medicine, University Hospital Hamburg; and the Institute of Pathology, University Hospital Hamburg, Hamburg, Germany.
- Address reprint requests to J.R. Izbicki, MD, PhD, Abt. für Allgemeinchirurgie, Universitätsklinikum Hamburg Eppendorf, Martinistrasse 52, D-20246 Hamburg, Germany; email: izbicki{at}uke .uni-hamburg.de.
Abstract
PURPOSE: Data on skip metastases and their significance are lacking for esophageal cancer. This issue is important to determine the extent of lymphadenectomy for esophageal resection. In this study we examined the lymphatic spread in esophageal cancer by routine histopathology and by immunohistochemistry.
PATIENTS AND METHODS: A total of 1,584 resected lymph nodes were obtained from 86 patients with resected esophageal carcinoma and evaluated by routine histopathology. Additionally, frozen tissue sections of 540 lymph nodes classified as tumor-free by routine histopathology were screened for micrometastases by immunohistochemistry with the monoclonal antibody Ber-EP4. The lymph nodes were mapped according to the mapping scheme of the American Thoracic Society modified by Casson et al.
RESULTS: Forty-four patients (51%) had pN1 disease, and 61 patients (71%) harbored lymphatic micrometastases detected by immunohistochemistry. Skip metastases detected by routine histopathology were present in 34% of pN1 patients. Skipping of micrometastases detected by immunohistochemistry was found in 66%. The presence of micrometastases was associated with a significantly decreased relapse-free and overall survival (56.0 v 10.0 months and > 64 v 15 months, P < .0001 and P = .004, respectively). Cox regression analysis revealed the independent prognostic influence of micrometastases in lymph nodes. Lymph node skipping had no significant independent prognostic influence on survival.
CONCLUSION: Histopathologically and immunohistochemically detectable skip metastases are a frequent event in esophageal cancer. Only extensive lymph node sampling, in conjunction with immunohistochemical evaluation, will lead to accurate staging. An improved staging system is essential for more individualized adjuvant therapy.
THE PROGNOSIS of patients with resectable esophageal carcinoma still remains poor. The reported 5-year survival rates range from 20% to 36% after intentionally curative surgery.1-3 Early metastatic relapse after complete resection of an apparently localized primary tumor4-6 indicates micrometastatic tumor cell spread at the time of surgery, undetected by current staging methods and by routine histopathology. Recently we were able to demonstrate a strong prognostic influence of individual tumor cells in lymph nodes identifiable by immunohistochemistry in patients with esophageal carcinoma, which were free of metastases on routine histopathologic examination.7
In view of the pivotal role of lymph node metastases in patients with esophageal cancer, it is essential not to miss any lymphatic involvement. Because discontinuous lymphatic spread occurs in some tumor entities,8,9 precise knowledge of the mode of lymphatic spread is imperative in patients with esophageal cancer to determine the extent of lymphadenectomy. The result of an adequate lymphadenectomy will not only have a prognostic impact but will also influence staging and adjuvant therapy. Thus far, the mode of lymphatic spread is not well described in esophageal cancer. Therefore, we conducted a prospective study to assess the exact mode of lymphatic tumor cell spread to lymph nodes and its prognostic significance. We were particularly interested in the pattern of micrometastatic tumor cell spread to lymph nodes as identified by immunohistochemistry.
PATIENTS AND METHODS
This study was approved by the ethics committee of the University of Hamburg. Informed consent was obtained from all patients before inclusion into the study. Lymph nodes and tumor samples were collected from 86 patients with resectable esophageal carcinoma who had undergone radical en bloc esophagectomy at our institution. All had tumor-free resection margins on microscopic examination of the surgical specimen. Reconstruction was performed with gastric tube interposition with a cervical esophagogastric anastomosis. Tumor stage and grade were classified according to the fourth edition of the International Union against Cancer10 by investigators unaware of the immunohistochemical findings. After primary surgery, patients were reexamined every 3 months for 2 years and every 6 months thereafter. The evaluation included physical examination, plain chest radiography, endoscopy, endoscopic ultrasound, computed tomography of the chest and the abdomen, abdominal ultrasound, tumor marker studies (squamous cell carcinoma antigen, carcinoembryonic antigen, and CA 19-9), and bone scanning. Six patients died within 90 days after surgery in the hospital as a result of complications (mortality rate, 6.9%). Complete follow-up was obtained from the remaining 80 patients.
Tissue Preparation and Histopathologic Analysis
Of 1,584 resected lymph nodes, 800 lymph nodes, judged to be tumor-free by the surgeon, were systematically sampled during lymphadenectomy from six different lymph node levels (periesophageal lymph nodes < 3 cm from the primary tumor, cervical lymph nodes, upper mediastinal lymph nodes, lower mediastinal lymph nodes, perigastric lymph nodes, and lymph nodes at the celiac trunk). No tumor cells were detected by routine histopathologic examination in 540 of 800 lymph nodes. All nodes were mapped by the surgeon according to the mapping scheme of the American Thoracic Society11 modified by Casson et al.12
All systematically sampled lymph nodes were divided into two parts. One part was embedded in paraffin for routine histopathologic staging (hematoxylin and eosin [HE]); the other part was snap-frozen in liquid nitrogen after removal and stored at −80°C until used. Lymph nodes without evidence of nodal metastases by routine histopathology were screened by immunohistochemistry, using the antiepithelial cell monoclonal antibody Ber-EP4 (immunoglobulin G [IgG1]; DAKO, Hamburg, Germany), which can be used on snap-frozen or paraffin-embedded material for the detection of isolated tumor cells as described previously.13 Ber-EP4 is an antibody against two glycopolypeptides of 34 and 49 kd on the surface of and in the cytoplasm of all epithelial cells (except for the superficial layers of squamous epithelia, parietal cells, and hepatocytes). The antibody does not react with mesenchymal tissue, including lymphoid tissue.14,15 From each lymph node, 6 to 8 μm cryostat sections were cut from three different levels and transferred on to glass slides pretreated with 3-triethoxysilyl-propylamine (Merck, Darmstadt, Germany). One section of each level was stained using the alkaline phosphatase–antialkaline phosphatase technique, as described.13 We have found consistently negative staining of lymph nodes from 24 control patients with nonepithelial tumors or inflammatory diseases.13 Sections of normal colonic mucosa served as positive staining controls, and isotype-matched, irrelevant murine monoclonal antibodies served as negative controls (MOPC 21 for IgG1; Sigma Chemical Co, St. Louis, MO).
All HE-stained and immunostained slides were evaluated in a blinded fashion by two observers independently. Both observers obtained identical results in 92% of the slides; the remaining slides were reevaluated and a consensus decision was made. Minimal tumor cell involvement in a lymph node that was considered tumor-free by routine histopathologic staining was defined as the presence of up to three Ber-EP4–positive cells within the body of the lymph node. In those patients who had Ber-EP4–positive cells in histopathologically “tumor-free” lymph nodes, two lymph node sections consecutive to the one containing the immunostained cells were prepared, stained with HE, and evaluated by a pathologist who had no knowledge of the initial results.
Evaluation of the Mode of Lymphatic Tumor Cell Spread
To assess the frequency of skip metastasis, we compared the lymphatic routes with respect to the site of the primary lesion using the mapping scheme according to the American Thoracic Society11 modified by Casson et al,12 and we divided the groups of lymph nodes into six levels ( Table 1): lymph nodes adjacent to the primary tumor, cervical nodes, upper mediastinal nodes, middle and lower mediastinal nodes, perigastric nodes, and nodes of the celiac trunk. If there was no continuous tumor cell spread from the primary tumor into the adjacent lymph node levels, missing one or more than one level was defined as skip metastasis. These skipped lymph node levels were tumor-free on routine histopathologic examination and/or immunohistochemical evaluation, respectively.
Lymph Node Mapping Scheme for Skip Micrometastases Detected by Immunohistochemistry
Statistical Analysis
Associations between categorical parameters were assessed with Fisher’s exact and χ2 tests. The Kaplan-Meier method was used to estimate overall cancer survival and relapse-free survival. For comparison purposes, log-rank tests were performed. Cox proportional hazards models were fitted for multivariate analysis.16 Differences between groups were considered significant if the P values were < .05 in a two-tailed test.
RESULTS
Patient Characteristics
For this study, specimens were examined from a total of 86 patients with resected esophageal carcinoma ( Table 2). Patients ranged in age from 34 to 76 years, and the median age was 58 years. Sixty-eight were male (79%) and 18 were female (21%). The primary lesion was located supracarinal in 50 (68%) and infracarinal in 36 patients (42%). Fifty-nine tumors were classified as squamous cell carcinomas (69%) and 27 as adenocarcinomas (31%). Lymph node involvement identified by routine histopathologic examination was found in 44 cases (51%).
Patient and Tumor Characteristics
Mode of Lymphatic Tumor Cell Spread Detected by Routine Histopathology
Skipping of one or more than one level of lymph node groups was present in 29 patients (34%) ( Table 3). Periesophageal lymph nodes (< 3 cm from the primary lesion) were skipped in 43% (n = 19) of pN1 patients. Skipping of the lower mediastinal lymph nodes and of the perigastric lymph nodes was observed in 30% (n = 13) and in 27% (n = 12), respectively. Fifty percent of the patients with infracarinal esophageal carcinomas (n = 18) had skip metastases compared with 26% of patients with supracarinal tumors (n = 11) (P = .01). Regarding tumor type, 31% of patients with squamous cell carcinoma (n = 18) and 41% of patients with adenocarcinoma (n = 11) showed skip metastases (not significant). The upper mediastinal lymph node group was skipped in one patient (2%).
Incidence of Skip Metastases
No correlation was found between the incidence of skip metastasis and tumor type, grade of differentiation, T stage, or N stage of the primary tumor.
Effect of Skip Metastasis on Survival
After a median observation time of 25 months (range, 1 to 67) the presence of skip metastasis was associated with a significantly decreased disease-free survival. The median relapse-free survival was 55 months for patients without skip metastasis, versus 10 months for patients with skip metastasis (P = .0031). Skip metastases were also of significant predictive value concerning overall survival. The median survival time of patients with skip metastasis was 15 months, compared with > 64 months for patients without skip metastasis (P = .0007). However, Cox regression analysis did not identify skip metastasis as an independent prognostic factor ( Table 5).
Multivariate Cox Regression Analyses for Relapse-Free Survival
Incidence of Micrometastatic Lymphatic Tumor Cell Spread Detected by Immunohistochemistry
Immunohistochemical analysis of cryostat sections of 540 histopathologically tumor-free lymph nodes from 86 patients showed single cells or cell clusters with positive Ber-EP4 staining in 97 nodes (18%) of 61 patients (71%). Twenty-six of 42 patients (62%) with disease staged as pN0 had nodal micrometastasis, compared with 35 of 44 patients (80%) with stage pN1 (not significant). The incidence of Ber-EP4–positive cells was independent from tumor localization, tumor type, grade of differentiation, T stage, or N stage.
Mode of Lymphatic Spread of Micrometastatic Cells
The mode of the micrometastatic cell spread into lymph nodes was also erratic (Table 3). Spatial progression of the Ber-EP4–positive cells through the regional node levels was observed in only 21 of 61 patients (34%). In the remaining group of patients, skipping of one or more lymph node levels by the tumor cells could be demonstrated. Twenty-three patients (46%) with supracarinal esophageal carcinomas showed skip metastases, compared with 17 patients (47%) with infracarinal tumors (not significant). In 10 of the patients with tumors located above the carina (39%), skip metastases were detected in lymph nodes of the celiac trunk.
In 18 of the Ber-EP4–positive patients (69%) with pN0 disease, skipping of Ber-EP4–positive cells was demonstrated. Thirteen of these patients (72%) showed skipping of the periesophageal lymph nodes < 3 cm from the primary lesion. The upper mediastinal lymph node group was skipped in 11% (n = 2). Skipping of the lower mediastinal lymph nodes and of the perigastric lymph nodes by Ber-EP4–positive cells was observed in 67% (n = 12) and in 39% (n = 7), respectively. There was no correlation between the incidence of skip metastases and tumor type, grade of differentiation, T stage, or N stage of the primary tumor.
Effect of Micrometastatic Cells on Survival
Patients with Ber-EP4–positive cells in their lymph nodes had a significantly decreased disease-free survival with 10.0 v 56.0 months for patients without Ber-EP4–positive cells (P = .0005) ( Table 4). Among the 37 patients with a histopathologic pN0 stage, 12 patients had disease without Ber-EP4–positive cells in their lymph nodes according to our immunohistochemical assay. Their median relapse-free survival period was 55.0 months, compared with 27 months for pN0 staged patients with Ber-EP4–positive cells (P = .0036). The presence of Ber-EP4–positive cells in lymph nodes was also of significant predictive value concerning overall survival. The median survival time of patients with a positive finding was 15.0 months, compared with > 64 months for patients with Ber-EP4-negative lymph nodes (P = .0044) (Table 4). There was no influence of skip metastasis in the patients with Ber-EP4–positive cells on relapse-free survival (without skip metastasis, n = 13 of 21; median, 7 months; and with skip metastases, n = 19 of 39; median, 11 months; P = .3055) and on overall survival (without skip metastases, n = 9 of 21; median, 24 months; and with skip metastases, n = 17 of 39; median, 15 months; P = .6488).
Prognostic Value of Ber-EP4–Positive Tumor Cells
Twenty-two pN1 patients (67%) had both skip metastasis detected by routine histopathology and skipping of micrometastasis detected by immunohistochemistry. Kaplan-Meier analysis did not reveal any significant prognostic influence of skip metastasis for this group of pN1 patients (data not shown).
Cox regression analysis revealed the independent prognostic influence of Ber-EP4–positive tumor cells in lymph nodes for relapse-free survival (hazard ratio [HR], 5.49; 95% CI, 1.76 to 17.08; P = .0033) and overall survival (HR, 3.88; 95% CI, 1.23 to 12.18, P = .0203), respectively. The grade of differentiation was also an independent prognostic factor for relapse-free survival. As expected also lymph node stage was a strong independent prognostic factor for relapse-free survival and overall survival (HR, 6.32 and 12.7, respectively) (Table 5).
DISCUSSION
Skipping of lymph nodes is present in a substantial number of patients with completely resected esophageal cancer. Remarkably, the mode of spread of micrometastatic cells is likewise erratic. Skip metastasis in lymph nodes is a well-described phenomenon in non–small cell lung cancer8 and papillary carcinoma of the thyroid.9 In the latter, skip metastases were present in one third of the studied cases and could be excised only by complete locoregional neck dissection, whereas in breast cancer skip metastases are a rare event.17,18
Pioneering work on the lymphatics of the esophagus was done in 1903 by Sakata,19 who revealed that the submucosal lymphatics do not drain segmentally, but in a longitudinal fashion. Lymph nodes that drain the esophagus are widely distributed because of the two different embryologic origins of the esophagus. Accordingly, the organ has two main lymphatic collection areas, in the abdomen around the celiac trunk and at the neck. Furthermore, extensive communications between the esophageal layers are present. These anatomic features of the esophageal lymphatics can result in an unpredictably vast tumor cell spread. Further explanations for skipping lymph nodes, especially in thoracic malignancies, are fibrosis and hyalinization of the lymphatic tissue due to preexisting anthracosilicosis, tuberculosis, or exposure to ionizing radiation. The skipping of structurally intact lymph nodes can be explained by anatomically demonstrable intra- and perinodal short circuit connections.20 Van de Ven et al21 reported an erratic spread of histopathologically verified lymphatic metastases in esophageal adenocarcinoma, and Akiyama et al22 showed positive nodes in the superior mediastinum in 9.8% of patients with primary tumors located in the lower esophagus and node metastases in the superior gastric area in 31.8% of patients with tumors located in the upper esophagus. However, no precise data regarding the exact mode of lymphatic spread in esophageal cancer, especially for minimal tumor cell spread, are available thus far.
To assess the frequency of skip metastasis of tumor cells identified by routine histopathology and of micrometastasis detected by immunohistochemistry, we compared the lymphatic routes with respect to the site of the primary lesion and the mapping scheme according to the American Thoracic Society11 modified by Casson et al12 and divided the groups of lymph nodes into six levels (Table 1). Skipping of one or more than one level was defined as skip metastasis. Interestingly, skipping of lymph nodes was more frequent in infracarinal tumors. However, this finding was independent of the tumor cell type.
Lymph node micrometastasis has been assessed in breast cancer histopathologically by numerous serial sections. However, this time-consuming method is not feasible in routine histopathology. Recently our group showed that immunohistochemical screening of sections from three different lymph node levels with the monoclonal antibody Ber-EP4 is a sensitive and specific assay for the detection of epithelial tumor cell spread.7,13 This method can possibly introduce a sampling error. The positive correlation of the survival analysis with the immunohistochemical analysis indicates, however, that evaluation of three sections might be sufficient.
The rather erratic mode of lymphatic spread in esophageal cancer supports the recommendation for an extensive lymph node sampling in combination with immunohistochemical evaluation for a better postoperative staging. The discussed concept of sentinel lymph node evaluation therefore seems not to be a feasible approach in esophageal cancer.
As demonstrated previously by our group,7 the presence of immunohistochemically identifiable micrometastatic tumor cells in lymph nodes is an independent prognostic factor in esophageal carcinoma patients. It is notable that all patients who had no nodal tumor involvement by both histopathologic and immunohistochemical analysis survived the median observation period of 25 months, whereas patients without lymphatic metastases on routine histopathology but with Ber-EP4–positive cells in their lymph nodes had an outcome similar to those with lymph node metastases proven by routine histopathology. Interestingly, the presence of immunohistochemically detected tumor cells in lymph nodes was the strongest independent indicator for relapse-free survival and overall survival in the statistical analysis.
Skip metastasis, detected by routine histopathology or by immunohistochemistry, had no independent prognostic impact. It seems that this phenomenon does not reflect a more aggressive or advanced disease. It is an indicator that limited lymph node sampling in the context of a complex lymphatic anatomy might lead to false-negative histopathologic staging.
In conclusion, the presence of Ber-EP4–positive tumor cells in lymph nodes is an independent prognostic factor for relapse-free and overall survival. Because the mode of tumor cell spread is erratic, we propose that the conjunction of extensive lymph node sampling and immunohistochemical examination of lymph nodes may be used to refine the current staging system for esophageal cancer. This might help to identify patients with a high risk of relapse despite a good prognosis according to conventional histopathology. Thus, immunohistochemical staining of lymph nodes could lead to a more individualized therapeutic concept for esophageal cancer patients with respect to adjuvant systemic treatment.
Acknowledgments
This work was supported by the Otto Foundation, Hamburg, and the Friedrich Bauer Foundation, Munich, Germany.
- Received July 17, 2000.
- Accepted December 21, 2000.
