Annals of Surgical Oncology Sign the Guestbook
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
10.1245/ASO.2004.04.018
Annals of Surgical Oncology 11:506-511 (2004)
© 2004 Society of Surgical Oncology
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Bollschweiler, E. H.
Right arrow Articles by Hölscher, A. H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Bollschweiler, E. H.
Right arrow Articles by Hölscher, A. H.

ORIGINAL ARTICLES

Artificial Neural Network for Prediction of Lymph Node Metastases in Gastric Cancer: A Phase II Diagnostic Study

Elfriede H. Bollschweiler, MD, Stefan P. Mönig, MD, Karin Hensler, MD, Stephan E. Baldus, MD, Keiichi Maruyama, MD and Arnulf H. Hölscher, MD, FACS, FRCS

From the Department of Visceral and Vascular Surgery, University of Cologne, Germany (EHB, SPM, KH, AHH); Institute of Pathology, University of Cologne, Germany (SEB); and Department of Surgical Oncology, University of Health and Welfare, Sanno Hospital, Tokyo, Japan (KM).

Correspondence: Address correspondence and reprint requests to: Elfriede Bollschweiler, MD, Klinik und Poliklinik für Visceral und Gefäßchirurgie, der Universität zu Köln, Joseph Stelzmann Straße 9, 50931 Köln, Germany; Fax: 49-221-478-5076; E-mail: elfriede.bollschweiler{at}medizin.uni-koeln.de


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 
Background: Extension of lymphadenectomy in gastric cancer is controversial, and preoperative diagnosis of lymph node metastases (LNM) is difficult. Therefore, knowledge-based systems such as the Maruyama computer program (MCP) are being developed. MCP shows good prognostic value for the compartments; however, for different lymph node groups (LNG) there are a large number of false positives. The aim of this study was to evaluate artificial neural networks (ANN) for predicting LNM in patients with gastric cancer and to compare the predictive power with that of MCP.

Methods: A total of 135 consecutive patients who underwent D2 gastrectomy were included. We applied a single-layer perceptron to the data of 4302 patients from the National Cancer Center, Tokyo, and compared the results with those from the MCP.

Results: Prediction of N+ or N0 with ANN-1 (Borrmann classification, T category, and tumor size and location) had an accuracy of 79%. The predictive value for LNM in each of the LNG varied: ANN-1, 64% to 86%; MCP, 42% to 70%. We constructed another ANN by using the additional parameter of metastases in LNG 3 as an example of sentinel node. The accuracy of this ANN was 93%.

Conclusions: Using an ANN, LNM in each LNG can be accurately predicted. Additional knowledge about one lymph node improves the results.

Key Words: Artificial neural network • Gastric cancer • Lymph node metastases • Sentinel lymph node • Diagnostic study


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 
The results of two prospective studies did not indicate how extensive a lymphadenectomy in gastric cancer should be in relation to the T category.1,2 The results of these studies showed higher mortality and morbidity rates after D2 lymphadenectomy than after D1 lymphadenectomy, and no difference seemed to exist in the long-term survival of patients after D1 or D2 lymphadenectomy.3,4 However, for some patients, radical lymph node (LN) dissection does improve prognosis. As a result of screening programs and of the widespread use of upper endoscopy during the last two decades, there has been an increased likelihood of resecting early-stage gastric cancer.5 The incidence of nodal involvement is reported to be as low as 2% when the depth of cancer invasion is limited to the mucosal layer, 18% when it is limited to the submucosal layer (T1), and approximately 50% if a carcinoma has invaded the muscular or subserosal layer (T2).6 Thus, it seems that a larger lymphadenectomy than necessary is performed in a considerably high proportion of patients with T1 and T2 cancer. Today, limited resections can be performed in patients with mucosa carcinoma. Prediction of the presence of LN metastases (LNM) is critical to accurately estimate the extent of lymphadenectomy needed. Imaging procedures look only to the size of LN; however, metastases can be in small LN, and conversely, large LN may be metastasis free.7

The lack of biological methods to predict the occurrence of metastasis before surgery has led to the search for methods that combine existing demographic data with mathematical models in an effort to better assess the presence of metastases and thus guide the surgeon in implementing appropriate surgical procedures. One such method is the Maruyama computer program (MCP), which calculates the probability of metastases in each LN group (LNG).8 The application of this program to German patients has shown good predictability for the compartments, but it results in a large number of false-positive predictions for the different LNGs.9 Similar results were reported recently by Guadagni et al.10

In the last 10 years a class of techniques inspired by the workings of biological neurons, artificial neural networks (ANN), have been proposed as a supplement or alternative to standard statistical techniques for predicting complex biological phenomena.11,12 Briefly, ANNs are a class of nonlinear mathematical models that are characterized by a complex structure of interconnected computational elements, the neurons. These computational elements aggregate a series of inputs (factors that influence the development of LNM) by using a summation operation and produce an output, such as the probability that an LN is infiltrated by tumor. The aim of this study was to evaluate a novel ANN for the prediction of LNM in patients with gastric cancer and to compare it with the results of the MCP.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 
Patients
All patients with gastric cancer who underwent total gastrectomy (n = 121) or subtotal gastrectomy (n = 14) at the Department of Surgery, University of Cologne, between May 1, 1996, and April 2001 were included in this study. Before surgery all patients underwent esophagogastroduodenoscopy with biopsy and histopathologic examination. In addition, endosonography of the stomach was performed to stage the depth of tumor infiltration (T category), and a computed tomographic scan of the chest, abdomen, and pelvis was performed to evaluate for metastases.

Surgical Procedure and Extent of Lymphadenectomy
In all cases, an en-bloc resection of the stomach with extended D2 lymphadenectomy was performed. The LN dissection included compartments I and II. Compartment I comprises all LNs along the lesser curvature (1, 3, and 5) and the greater curvature (2, 4, and 6) of the stomach. Compartment II comprises LN stations 7 to 12 according to the Japanese classification of gastric carcinoma.13 Type II (cardial) and type III (subcardial) adenocarcinomas of the gastroesophageal junction, according to the Siewert et al. classification system,14 were treated with a transhiatal extended gastrectomy including D2 lymphadenectomy and LN dissection of the lower mediastinum. In case of subtotal gastrectomy (16 patients with cT1 or cT2 tumor located in the antrum), only LN stations 3 to 6 (compartment I) and LN stations 7 to 12 (compartment II) were resected en bloc. Sampling of compartment III and IV nodes (13 to 16) was optional. The surgeon divided the en bloc–resected tissue containing LNs into separate stations and assigned numbers to these stations according to the Japanese classification system.13 A splenectomy was performed (n = 35) in cases of proximal gastric carcinoma (type II and III) and in cases of metastatic involvement of the splenic hilus nodes (station 10), but not in general.

Histopathologic Evaluation of Gastric Cancer LNs
Immediately after surgery, the regional LN stations adhering to gastrectomy specimens were separated in consultation with pathologists. Regional or distant LN stations were resected separately. All tissues were fixed in 5% neutral buffered formaldehyde for at least 18 hours at room temperature. After fixation, the LNs were prepared for histological examination. Briefly, by using an automated system (VIP; Vogel, Giessen, Germany), the tissues were dehydrated by sequential exposure to a graded alcohol series from 70% to absolute ethanol and finally xylene. Infiltration and embedding of dehydrated specimens were performed with paraffin (Paraplast; Sherwood, St. Louis, MM) that melted at a low temperature (56°C). Four sections (2 mm) were cut from each paraffin block and attached to two slides; one slide was stained with hematoxylin and eosin and the other by the periodic acid–Schiff reaction.

Artificial Neural Network
The methodology of our neural network has been described in detail elsewhere.15 In brief, we used a simple variant of feed-forward neural networking—the perceptron.16 It consists of two layers of so-called neurons—an input layer and an output layer—with one layer of synapses between them. The output layer includes two neurons representing metastasis and no metastasis of the LNG (Fig. 1). According to the activity of the input neurons and the weights of their synapses, the activities of the output neurons are calculated (processing). The process is similar to the mechanism of information processing in the brain: multiple synapses lead information from many nerve cells to one neuron. Incoming impulses are added up and transformed into neuron activity, which is represented by the frequency of leading impulses through the axon to other neurons.



View larger version (35K):
[in this window]
[in a new window]
  		FIG. 1. Example of an artificial neural network for preoperative prediction of lymph node group (LNG) 13 by using the input variables T category (endosonography), Borrmann classification, location, and maximal diameter of the tumor (endoscopy).

 
For this study, a data set with 4302 records from the National Cancer Center, Tokyo, encompassing the years 1969 to 1988 was available to us. This data set included information about patients’ age and sex; size, site, and depth of invasion of the tumor; Borrmann classification17; histology of the tumor; and variables for the number of affected LNs in each of the 16 LNGs. The number of LNGs was classified according to the Japanese classification of gastric carcinoma.13 One third of the data were used to train the ANN, one third to improve it, and one third for testing the system. At least four variables demonstrated predictive importance for LNM: depth of invasion, maximal diameter, Borrmann classification, and location of the tumor. Other variables, such as age, sex, or histology, had low or no bearing on the prediction of LNM. The receiver operating characteristic curve was used to discover the best cutoff value.15

Data Collection and Statistics
The results of the preoperative staging were documented immediately after the diagnostic procedure. The neural network and the MCP were used prospectively for each patient. The postoperative results of pathology were documented in a standardized manner. Calculations of the accuracy, sensitivity, specificity, positive predictive value, and negative predictive value were based on the following definitions.

  1. Accuracy = (true positive + true negative)/all patients.
  2. Sensitivity = true positive/(true positive + false negative).
  3. Specificity = true negative/(true negative + false positive).
  4. Positive predictive value = true positive/(true positive + false positive).
  5. Negative predictive value = true negative/(true negative + false negative).

The output of the MCP is based on group means of patients with same background. Sensitivity and specificity were evaluated at various thresholds between probability values of 0.1% and 20%. A threshold of 7% above the LNG was classified as affected; errors in both directions were nearly similar. All calculations were performed with SPSS for Windows 9.0 (SPSS Inc., Chicago, IL).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSION
 REFERENCES
 
A total of 135 patients with gastric cancer were operated on in the department of surgery. The epidemiological data are shown in Table 1. The mean number of resected LNs was 37.7. The number of metastatic LNs was on average 7.6 (minimum, 0; maximum, 48).


View this table:
[in this window]
[in a new window]
  		TABLE 1. Clinicopathologic data of 135 patients with gastric cancer
 
The accuracy of prediction with a separate ANN (ANN-1) for each LNG varied from 64% to 86%. The accuracy of ANN-1 was, on average, 15% better than the prediction with the MCP (Table 2). The accuracy for compartment I was 78% with ANN-1 and 62% with MCP, and for compartment II, it was 79% with ANN-1 and 66% with MCP. We did not calculate the values for LNGs 10, 14, 15, and 16 because metastatic LNs were found in only few cases, and these LNGs were resected only in special cases.


View this table:
[in this window]
[in a new window]
  		TABLE 2. ANN-1 and MCP
 
Table 3 shows the prediction of N+ with ANN-1 by using the following variables: depth of tumor infiltration, size and location of the tumor, and Borrmann classification. The accuracy of ANN-1 was 79%, the sensitivity 88%, and the specificity was 55%; the positive and negative predictive value was 77%, respectively.


View this table:
[in this window]
[in a new window]
  		TABLE 3. ANN-1a
 
To improve the accuracy of the prediction of N+ or N0, we constructed another neural network (ANN-2) by using the same parameters as in ANN-1 with additional information about metastases in LNG 3. LNG 3 was chosen as an example of a positive sentinel LN. The accuracy of prediction with a separate ANN (ANN-2) for each LNG (without LNG 3) varied from 78% to 100%. The accuracy of ANN-2 was, on average, 13% better than the prediction with ANN-1. The results are shown in Table 4.


View this table:
[in this window]
[in a new window]
  		TABLE 4. ANN-2a
 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSION
 REFERENCES
 
LNMs are the most important prognostic factors in patients with gastric cancer.3,18–22 Staging of LNM is important not only for the choice of surgical techniques or postoperative tumor staging, but also for accurate preoperative staging and therapeutic decisions, including endoscopic mucosa resection or neoadjuvant treatment. Until now, no method has been available for correct preoperative staging of N category in gastric cancer patients.23 Even the accuracy of N staging by computed tomography or endoscopic ultrasonography, if not limited to regional LNGs, is unsatisfactory.24–26

A novel approach to predict LN involvement is computer programs for pretherapeutic staging. Maruyama from the National Cancer Center, Tokyo, developed a computer program for evaluating survival time and infiltration of LNs in individual cases.8 The MCP was evaluated in a cohort of German patients with gastric cancer to examine the validity of its predictive power in this population.9 The program was accurate in its prediction for the incidence of metastasis in compartments II (89%) and III (96%); however, the predictive power for compartment I was lower, namely, 82%. In a study of a cohort of Italian patients, the accuracy of LNM prediction was 72.4% for stations 13 to 16, 81.6% for stations 7 to 12, and 83.4% for stations 1 to 6. The false-positive rate was similar to that of the German patients, but the negative rate was higher.10 The results of our study with the MCP show lower accuracy than the other two studies. This may be explained by a high rate—50% in this study—of tumors in the upper third of the stomach compared with the Italian study, in which only 24% of patients had tumors in the upper third of the stomach, and the Japanese study, in which only 19% of the patients had tumors in the upper third of the stomach. The computer program is based on the Japanese data.

Neural networks can play a critical role in medical decision support because they are effective in multifactorial analysis. More specifically, neural networks can use multiple factors in resolving a medical diagnosis whose classification is based on multiple factors, thus reducing the error in the diagnosis for a population of patients.

The results of our study demonstrate that ANNs were able to predict LNM in each LNG of gastric cancer with an accuracy that was, on average, 15% better than the prediction with the MCP. Similar results were obtained with the construction of a neural network for the prediction of the N category (N+ or N0). Data of an earlier study showed that other statistical methods, such as logistic regression analysis, would lead to systematic overstaging of N category, with a median sensitivity of 0.0%.15

During the construction of the neural network, we became aware that results with neural networks always depend on the data with which they were trained. Neural networks are excellent in identifying and learning patterns that are present in the data; however, if a neural network is trained to predict a medical outcome, then there must be predictive factors among the inputs to the network before the network can learn to perform this prediction successfully.27 For the development of our ANN, we used the data of the National Cancer Center in Tokyo, because no significant differences were found in a study comparing survival rates and LN involvement between German and Japanese patients.28 The conditions for the input variables of our networks were that they had to exist before surgery and that they had to contain predictive factors for LNM. All variables of the MCP were examined in a multivariate analysis.29 However, it is obvious that the variables used in our various neural networks were not sufficient to predict metastasis with high accuracy.

To improve the accuracy of ANN-1, we trained a neural network by adding one additional factor: information about metastases in LNG 3. With this additional information, the predictive accuracy was improved to 93%, with a negative predictive value of 95% and a positive predictive value of 92%.

In our study, we used knowledge of pathologic results from LNG 3 to construct another neural network. This additional information may be obtained from the sentinel LN biopsy. During the last 2 years, the lymphatic drainage system of the stomach has been studied by using radiography with various dyes, and several major routes have been identified.30–35 Gastric lymphatic channels are multidirectional and form complex networks. Sano et al.32 were able to show that the perigastric nodal area close to the primary tumor was the first site of metastasis in 62% of the cases. It is interesting to note that the contrast media does not stain all nodes on the network, and most of it skips the metastases that occur by such multibypass streams.34 Under these conditions, it is impossible to select several nodes from black-stained nodes for the purpose of evaluating extension of metastases. Therefore, it would be of interest to combine our neural network with the information from sentinel LN biopsy.

Our group has identified protein markers as predictors of LNM36–38; specifically, the immunoreactivity of mmp2, p53, and p21 was correlated with the risk of developing LNM in gastric carcinoma. The combination of ANN and these protein markers could improve the accuracy for predicting LNM before surgery.


   CONCLUSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSION
REFERENCES
 
By using an ANN, LNM in each of the 16 LNG, according to the Japanese classification, can be predicted with acceptable accuracy. In addition, we constructed a neural network by using the information of the metastatic behavior of one LNG to diagnose the N category before surgery. This neural network predicted N+ or N0 with an accuracy of 93%.


   FOOTNOTES
 
In a prospective diagnostic phase II study, neural networks could predict metastases with good accuracy for each lymph node group in gastric cancer. By including the information of a specific sentinel node, the accuracy of the neural network was improved.

Received for publication April 14, 2003. Accepted for publication January 6, 2004.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 

  1. Bonenkamp JJ, Songun I, Hermans J, et al. Randomised comparison of morbidity after D1 and D2 dissection for gastric cancer in 996 Dutch patients. Lancet 1995; 345: 745–8.[CrossRef][Medline]
  2. Cuschieri A, Fayers P, Fielding J, et al. Postoperative morbidity and mortality after D1 and D2 resections for gastric cancer: preliminary results of the MRC randomised controlled surgical trial. Lancet 1996; 347: 995–9.[CrossRef][Medline]
  3. Bonenkamp JJ, Hermans J, Sasako M, van der Velde CJH. Dutch Gastric Cancer Group: extended lymph-node dissection for gastric cancer. N Engl J Med 1999; 340: 908–12.[Abstract/Free Full Text]
  4. Cuschieri A, Weeden S, Fielding J, et al. Patients survival after D1 and D2 resections for gastric cancer: longterm results of MRC randomized surgical trial. Br J Cancer 1999; 79: 1522–8.[CrossRef][Medline]
  5. Perri F, Iuliano R, Valente G, et al. Minute and small gastric cancers in a Western population: a clinicopathologic study. Gastrointest Endosc 1995; 41: 475–80.[CrossRef][Medline]
  6. Sasako M, Mc Culloch P, Kinoshita T, Maruyama K. New method to evaluate the therapeutic value of lymph node dissection for gastric cancer. Br J Surg 1995; 82: 346–51.[Medline]
  7. Mönig S, Zirbes TK, Schröder W, et al. Staging of gastric carcinoma—correlation of lymph node size and metastatic infiltration. Am J Roentgenol 1999; 173: 365–7.[Abstract/Free Full Text]
  8. Kampschöer GHM, Maruyama K, van de Velde CJH, Sasako M, Kinoshita T, Okabayashi K. Computer analysis in making preoperative decisions: a rational approach to lymph node dissection in gastric cancer patients. Br J Surg 1989; 76: 905–8.[Medline]
  9. Bollschweiler E, Boettcher K, Hoelscher AH, et al. Preoperative assessment of lymph node metastases in patients with gastric cancer: evaluation of the Maruyama computer program. Br J Surg 1992; 79: 156–60.[Medline]
  10. Guadagni S, de Manzoni G, Catarci M, et al. Evaluation of the Maruyama computer program for preoperative estimation of lymph node metastases from gastric cancer. World J Surg 2000; 24: 1550–8.[CrossRef][Medline]
  11. Sargent DJ. Comparison of artificial neural networks with other statistical approaches—results from medical data sets. Cancer 2001; 91 (8 Suppl): 1636–42.[CrossRef][Medline]
  12. Dvorchik I, Subotin M, Marsh W, McMichael J, Fung JJ. Performance of multilayer feedforward neural networks to predict liver transplantation outcome. Methods Inf Med 1996; 35: 12–8.[Medline]
  13. Japanese Gastric Cancer Association. Japanese classification of gastric carcinoma. 2nd English edition. Gastric Cancer 1998; 1: 10–24.[Medline]
  14. Siewert JR, Hölscher AH, Becker K, Gössner W. Kardiakarzinom. Versuch einer therapeutisch relevanten Klassifikation. Chirurg 1987; 58: 25–9.[Medline]
  15. Droste K, Bollschweiler E, Waschulzik T, et al. Prediction of lymph node metastasis in gastric cancer patients with neural networks. Cancer Lett 1996; 109: 141–8.[CrossRef][Medline]
  16. Rosenblatt F. The perceptron: a probabilistic model for information storage and organisation in the brain. Psychol Rev 1958; 65: 386–408.[CrossRef][Medline]
  17. Borrmann R. Geschwulste des Magens. In: Henke FU, Lubarsch O, eds. Handbuch der Speziellen Pathologischen Anatomie und Histologie. Berlin: Springer Verlag, 1926.
  18. Sendler A, Dittler AJ, Feussner H, et al. Preoperative staging of gastric cancer as precondition for multimodal treatment. World J Surg 1995; 19: 501–8.[CrossRef][Medline]
  19. Siewert JR, Böttcher K, Roder JD, et al. Prognostic relevance of systematic lymph node dissection in gastric carcinoma. Br J Surg 1993; 80: 1015–8.[Medline]
  20. Schröder W, Hölscher AH. Aktuelle Therapiestrategien beim Magenfrühkarzinom. Onkologe 2001; 7: 610–22.[CrossRef]
  21. De Manzoni G, Verlato G, Guglielmi A, Laterza E, Genna M, Cordiano C. Prognostic significance of lymph node dissection in gastric cancer. Br J Surg 1996; 83: 1604–8.[Medline]
  22. Siewert JR, Böttcher K, Stein HJ, Roder JD. Relevant prognostic factors in gastric cancer: ten-year results of the German Gastric Cancer Study. Ann Surg 1998; 288: 449–61.
  23. Bollschweiler E, Mönig S, Hölscher AH. Lymphknotenmetastasierung—kann man sie vorhersagen? Onkologe 2001; 7: 604–9.[CrossRef]
  24. Akahoshi K, Chijiwa Y, Hamada S, et al. Pretreatment staging of endoscopically early gastric cancer with a 15 MHz ultrasound catheter probe. Gastrointest Endosc 1998; 48: 470–6.[CrossRef][Medline]
  25. Davies J, Chalmers AG, Sue-Ling HM, et al. Spiral computed tomography and operative staging of gastric carcinoma: a comparison with histopathological staging. Gut 1997; 41: 314–9.[Abstract/Free Full Text]
  26. Kim AY, Han JK, Seong CK, Kim TK, Choi BI. MRI in staging advanced gastric cancer: is it useful compared with spiral CT? J Comput Assist Tomogr 2000; 24: 3389–94.
  27. Dayhoff JE, DeLeo JM. Artificial neural networks—opening the black box. Cancer 2001; 91: 1615–35.[CrossRef][Medline]
  28. Bollschweiler E, Böttcher K, Hölscher AH, et al. Is the prognosis for Japanese and German patients with gastric cancer really different? Cancer 1993; 7: 2918–25.
  29. Maruyama K. The most important prognostic factors for gastric cancer patients: a study using univariate and multivariate analysis. Scand J Gastroenterol 1987; 22 (Suppl 133): 3–68.
  30. Ichikura T, Morita D, Uchida T, et al. Sentinel node concept in gastric carcinoma. World J Surg 2002; 26: 318–22.[CrossRef][Medline]
  31. Hiratsuka M, Miyashiro I, Ishikawa O, et al. Application of sentinel node biopsy to gastric cancer surgery. Surgery 2001; 129: 335–40.[CrossRef][Medline]
  32. Sano T, Katai H, Sasako M, Maruyama K. Gastric lymphography and detection of sentinel nodes. Recent Results Cancer Res 2000; 157: 253–8.[Medline]
  33. Kitagawa Y, Ohgami M, Fujii H, et al. Laparoscopic detection of sentinel lymph nodes in gastrointestinal cancer: a novel and minimally invasive approach. Ann Surg Oncol 2001; 8: 86S–89S.[Medline]
  34. Maruyama K, Sasako M, Kinoshita T, Sano T, Katai H. Can sentinel node biopsy indicate rational extent of lymphadenectomy in gastric cancer surgery? Langenbecks Arch Surg 1999; 384: 149–57.[CrossRef][Medline]
  35. Aikou T, Higashi H, Natsugoe S, Hokita S, Baba M, Tako S. Can sentinel node navigation surgery reduce the extent of lymph node dissection in gastric cancer? Ann Surg Oncol 2001; 8 (9 Suppl): 90–3.
  36. Mönig SP, Baldus SE, Hennecken JK, et al. Expression of MMP-2 is associated with progression and lymph node metastasis of gastric carcinoma. Histopathology 2001; 39: 597–602.[CrossRef][Medline]
  37. Baldus SE, Schneider PM, Mönig SP, et al. p21 in gastric cancer: associations with histopathological subtypes. Lymphnodal metastasis, prognosis and p53 status. Scand J Gastroenterol 2001; 36: 975–98.[Medline]
  38. Mönig SP, Eidt S, Zirbes TK, Stippel D, Baldus SE, Pichlmaier H. p53 expression in gastric cancer: clinicopathological correlation and prognostic significance. Dig Dis Sci 1997; 42: 2463–7.[CrossRef][Medline]



This article has been cited by other articles:


Home page
 Ann. Surg. Oncol.Home page
L. F. Onate-Ocana, V. Aiello-Crocifoglio, D. Gallardo-Rincon, R. Herrera-Goepfert, R. Brom-Valladares, J. F. Carrillo, E. Cervera, and A. Mohar-Betancourt
Serum Albumin as a Significant Prognostic Factor for Patients with Gastric Carcinoma
Ann. Surg. Oncol., February 1, 2007; 14(2): 381 - 389.
[Abstract] [Full Text] [PDF]

Home page
 JPEN J Parenter Enteral NutrHome page
R. N. Dickerson, D. L. Mason, M. A. Croce, G. Minard, and R. O. Brown
Evaluation of an Artificial Neural Network to Predict Urea Nitrogen Appearance for Critically Ill Multiple-Trauma Patients
JPEN J Parenter Enteral Nutr, November 1, 2005; 29(6): 429 - 435.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Bollschweiler, E. H.
Right arrow Articles by Hölscher, A. H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Bollschweiler, E. H.
Right arrow Articles by Hölscher, A. H.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS