Long-term results of early gastric cancer accomplished in a European institution by Japanese-type radical resection

Transcription

1 A. Gastric Pertl Cancer et al.: Long-term (1999) 2: results of early gastric cancer by International and Japanese Gastric Cancer Associations Original article Long-term results of early gastric cancer accomplished in a European institution by Japanese-type radical resection Alexander Pertl 1, Michael Jagoditsch 1, Gerhard R. Jatzko 1, Helmut Denk 2, and Haro M. Stettner 3 1 Surgical Department, Hospital of Barmherzige Brüder, Teaching Hospital of the Faculty of Medicine, University of Vienna, Spitalgasse 26, A-9300, St. Veit/Glan, Austria 2 Department of Pathological Anatomy, University of Graz, Austria 3 Department of Mathematics and Applied Statistics, University of Klagenfurt, Austria Abstract: Background. Long-term survival following Japanese-type radical surgery for 130 consecutively performed early gastric cancers (EGC) in a single Austrian institution between January 1, 1984 and May 31, 1998 was analyzed in terms of longterm survival, postoperative morbidity, and mortality. Methods. Extended D2 lymphadenectomy as defined by the JRSGC was performed in 129 patients with EGC. The surgical process was consistent as nearly all patients were operated on by only two surgeons. Overall survival and factors influencing survival were analyzed with particular regard to the depth of tumor infiltration, histological type, tumor grading, Lauren classification, tumor diameter, macroscopic appearance, localization, and lymph node involvement. Results. Of 678 gastric cancer patients surgically treated in the mentioned period, 130 patients (19.2%) were qualified as EGC. In 70 patients the tumor was limited to the mucosa and in 60 patients the tumor had not yet invaded the submucosa. The percentage of patients with positive lymph nodes increased from 2.9% with mucosal invasion to 21.7% with submucosal tumor involvement. The overall 5- and 10-year observed survival rate, postoperative mortality not excluded, was 74.6% and 62.1%, respectively, and 91.1% and 91.1%, respectively, when calculated as tumor specific. The 5- and 10- year observed survival rate of tumors limited to the mucosa was 77.2% and 72.1%, respectively, and 98.1% and 98.1%, respectively, when calculated as tumor specific. The respective values for submucosal invasion were 71.6% and 51.7%, and 82.7% and 82.7%. Postoperative complications occurred in 17 patients (13.1%) and postoperative hospital mortality totaled 1.5% (2/130). In multivariate analysis, only lymph node metastases were found to have independent prognostic influence on survival (P < 0.001; hazard ratio, 8.25). Conclusion. Japanese-type radical lymph node dissection for EGC in a European surgical institution yielded long-term survival nearly identical to that reported repeatedly by Japanese authors. Postoperative morbidity and mortality was not sacrificed by our comparatively radical surgical approach. Offprint requests to: G.R. Jatzko Received for publication on March 8, 1999; accepted on June 7, 1999 Key words: early gastric cancer, Japanese-type radical resection, long-term results Introduction The term early gastric cancer, or EGC, was introduced in 1962 by the Japanese Gastroenterological Endoscopic Society to describe gastric cancer limited to the mucosa or submucosa with resulting favorable prognosis when treated exclusively with surgical therapy [1 3]. Indeed, 5-year survival rates up to 95%, especially in Japan, have been reported repeatedly [3 11]. There is, however, a distinct difference between EGC of the mucosa and submucosa types [5 11]. Pathohistological studies in Japan have identified criteria for subtypes of mucosal EGC with minimal probability for lymph node metastases [9,12 16]. This characterization resulted in the introduction of endoscopical therapy modalities for these EGC types of low malignancy, which has been confirmed by long-term results [13,15 17]. For all other EGC, radical lymph node dissection according to the JRSGC was suggested by Japanese institutions [18 24]. Because of the high incidence of gastric cancer in Japan, extensive screening programs could have been established there decades ago [9,25]. EGC frequency nowdays is reported to be 40% 60% [5,9,26], whereas the incidence in the United States or Europe ranges between 4% and 16% [27 32]. Survival of 90% and more has been reported by Japanese authors [3 11], whereas survival in Europe is somewhat lower [29,30]. There is still much discrepancy related to radical lymph node dissection for prognosis, not only for EGC but also for advanced gastric cancer, between Japanese and most Western institutions, unconvincingly suggesting differences in tumor biological behavior [33,34]. This study presents a relatively large collection of EGC from a single European institution treated by

2 116 A. Pertl et al.: Long-term results of early gastric cancer means of Japanese-type lymph node dissection, which resulted in a similar or nearly identical outcome for the patients. Patients and methods Between January 1, 1984 and May 31, 1998, 678 patients were admitted with gastric cancer. In 130 patients (19.2%) the tumor did not invade beyond the submucosal layer and therefore was qualified as EGC as defined by the Japanese Society for Gastroenterological Endoscopy. These patients are the subjects of this analysis. The median age was 66.5 years (range, 34 85); 71 (54.6%) patients were men and 59 (45.4%) were women. The criteria defined by the WHO [35] and the UICC [36 38] were used for pathological classification purposes, with particular regard to the pt-1 category. The Lauren classification was used for histological typing [39]. Hospital records including pathology, surgery, patients charts, and information about follow-up status were critically examined. Follow-up was either carried out at our department or by some specialized gastroenterologists in cooperation with the hospital. The mean observation time was (SE 121.2) days, 67.1 (SE 4.0) months, respectively. Only 7 patients were lost to follow-up after a mean observation time of 93.9 (SE 16.4) months. The extent of resection, subtotal or total gastrectomy, depended on tumor localization, presence of multicentric lesions, and the histological classification according to Lauren. Subtotal gastrectomy was accomplished in most cases by Billroth II type resection and in exceptional cases by Billroth I type resection; gastric replacement with Roux-en-Y esophagojejunostomy was the standard procedure after total gastrectomy. The spleen was removed en bloc in all cases of total gastrectomy. Obligatory D2 lymph node dissection with additional removal of station 13 was carried out independent of the type of resection according to the recommendations of the JRSGC [18,19]. Obligatory dissection included the regional perigastric lymph nodes along the lesser and greater curvatures, pericardial lymph nodes, and nodes located along the left gastric, common hepatic, splenic, and coeliac arteries, as well as lymph nodes along the hepatoduodenal ligament and those of the posterior aspect of the head of the pancreas. Statistical analysis Data were analyzed using the EGRET software package (Statistics and Epidemiology Research Corporation, Seattle, WA, USA); percentages were compared with the chi-square test, Fisher Yates test and Mann Whitney U-test, and survival rates were calculated according to the Kaplan Meier model [40]. Prognostic variables, such as depth of tumor infiltration, histological type, Lauren classification, tumor diameter, macroscopic appearance, localization, and lymph node involvement were initially analyzed with the log-rank test [41] and then entered into the Cox proportional hazards regression models [42] to identify factors with independent prognostic influence on survival. The stability of the model was guaranteed by using a stepbackward and step-forward fitting procedure. In the step-backward approach, the least significant variable was repeatedly eliminated after the initial simultaneous analysis until a core of variables with independent influence remained. In the step-forward approach, factors were entered into multivariate analysis according to their significance in the log-rank test. The next most significant factor was always added to the model and nonsignificant factors (P 0.05) were eliminated. The variables identified to have independent influence on survival were identical, regardless of the fitting procedure. Results Epigastric complaints led to the diagnosis of EGC in 89 patients (68.5%); 6 patients (4.6%) had a history of ulcer disease, but 35 patients (26.9%) had no specific complaints and thus the diagnosis of these patients was made by chance, on the occasion of a general examination. Tumor resection was possible in all 130 cases. Subtotal gastrectomies were performed in 89 (68.5%) and total gastrectomies in 40 (30.8%) of patients. A local excision was performed for only 1 patient with advanced hepatic cirrhosis. The median number of lymph nodes removed amounted to 21 (range, 3 60). Over the reported period, the majority of lesions, n 47 (56.9%), were located in the distal stomach third; 44 (33.8%) of EGC were found in the middle third, and 12 (9.2%) were located in the upper gastric third. Multicentricity of lesions was found in 9 patients (6.9%). No statistically significant correlations were obvious for tumor location and the incidence of lymph node metastases. The distribution of macroscopical appearance is presented in Table 1. The median diameter of lesions totaled 20mm (range, 2 140). EGC limited to the mucosa had a median diameter of 20mm (range, 2 100), and those with submucosal invasion of 20mm (range, 3 140). The difference in size was not statistically significant. Patients with intestinal types of EGC were significantly older, with a median age of 69 years (range, 37 85), than patients with diffuse or mixed types, who

3 A. Pertl et al.: Long-term results of early gastric cancer 117 Table 1. Macroscopic classification and correlation to lymph node involvement and tumor penetration Lymph node Mucosa Submucosa Type Number (%) metastases (%) type (%) type (%) Protruded type I 27 (20.8) 3 (11.1) 15 (55.6) 12 (44.4) Superficial Elevated type IIa 16 (12.3) 1 (6.3) 7 (43.8) 9 (56.2) Flat type IIb 25 (19.3) 1 (4.0) 12 (48.0) 13 (52.0) Superficial Depressed type IIc 35 (26.9) 4 (11.4) 20 (57.1) 15 (42.9) Excavated type 22 (16.9) 5 (22.7) 13 (59.1) 9 (40.9) Not classified 5 (3.8) Table 2a. Lymph node involvement according to the Lauren classification Lymph node Lauren type n (%) positive, n (%) pn1 (%) pn2 (%) pn3 (%) Intestinal 79 (60.8) 5 (6.4) 4 (5.1) 1 (1.3) Diffuse 36 (27.7) 7 (19.4) 3 (8.3) 4 (11.1) Mixed 15 (11.5) 3 (20) 3 (20) Table 2b. Lymph node involvement according to depth of penetration Lymph node Type n (%) positive, n (%) pn1 (%) pn2 (%) pn3 (%) Mucosal 70 (53.8) 2 (2.9) 2 (2.9) Submucosal 60 (46.2) 13 (21.7) 8 (13.3) 5 (8.3) were of median age of 58 years (range, 34 83) (P 0.001). The majority of lesions, n 79 (60.8%), was of intestinal type according to Lauren, whereas 51 (39.2%) were classified as diffuse or mixed types according to the Lauren classification. There was a correlation between Lauren classification and the incidence of lymph node metastases (P 0.022) in our series (Table 2a). Pathohistological examinations showed 70 (53.8%) EGC limited to the mucosa and 60 (46.2%) invading the submucosa (Table 2b). Lymph node metastases were found in only 2 patients (2.9%) with EGC limited to the mucosa, whereas 13 of patients (21.7%) with EGC invading the submucosa already had lymph node metastases (P 0.001) (Table 2b). Five cases of EGC (Table 1) with positive nodes (or 35.7% of all lymph node-positive patients) had to be assigned macroscopically to type III. Six adenocarcinomas (4.6%) were highly differentiated, 63 (48.5%) moderately well, and 61 (46.9%) poorly differentiated; 16 (12.3%) were signet-ring cell carcinomas, according to the WHO classification [35]. There was a statistically significant correlation between tumor grade (G1 G4) (P 0.028) and the incidence of positive nodes. No statistical significant correlation was observed for tumor grading and depth of tumor infiltration. Tumor diameter less than or greater than 3 cm showed no positive correlation in terms of lymph node metastases. Another primary malignancy was detected in only 1 patient; a 49-year-old man had concurrent carcinomas of the colon and jejunum. Postoperative complications were observed in 17 (13.1%) patients, and hospital postoperative mortality totaled 1.5% (2/130). There was no difference in postoperative mortality between patients undergoing subtotal gastrectomies (2.2%) and patients with total gastrectomies (0%) (P 0.474). There was also no correlation in postoperative mortality between patients younger and older than 65 years, with a postoperative mortality of 1.7% versus 1.4% (P 0.704). Postoperative complications, relaparotomies, ultrasound-guided drainages, and postoperative death are presented in Table 3. Five- and 10-year observed survival rates for all EGC patients, postoperative deaths not excluded, were

4 118 A. Pertl et al.: Long-term results of early gastric cancer Table 3. Postoperative complications, resulting relaparotomies, ultrasound-guided drainages, and deaths Ultrasound-guided Relaparotomies drainages Deaths Complications n n (%) n (%) n (%) Subphrenic abscess 8 1 (12.5) 5 (62.5) Anastomotic leakage 1 1 (100) 1 (100) Pancreatic fistula 2 1 (50) Intestinal obstruction 2 2 (100) Cardiac insufficiency 1 Pulmonary embolism 1 1 (100) Wound infection 1 Hemorrhage 1 1 (100) Total 17 6 (35.3) 5 (29.4) 2 (11.8) Fig. 1. Five- and 10-year observed (dashed line) and tumor-specific (solid line) survival probability for 130 patients with early gastric cancer (EGC) 74.6% (SE 4.4%) and 62.1% (SE 5.8%), respectively. Tumor-specific 5- and 10-year survival amounted to 91.1% (SE 2.9%) and 91.1% (SE 2.9%). For patients with mucosal EGC, the observed rates were 77.3% (SE 5.9%) and 72.1% (SE 6.6%), respectively, or 98.1% (SE 1.9%) and 98.1% (SE 1.9%) when calculated as tumor specific. Submucosal EGC showed 5- and 10-year observed survival of 71.6% (SE 6.4%) and 51.7% (SE 8.9%), respectively, and 82.7% (SE 5.8%) and 82.7% (SE 5.8%) when evaluated as tumor specific. Figure 1 shows overall observed and tumor-specific survival for EGC patients calculated according to the Kaplan Meier method; Fig. 2 shows tumor-specific survival rates for mucosal and submucosal lesions. Recurrences were observed in ten patients (7.7%), six of whom had positive nodes; four patients (40%) showed disseminated peritoneal recurrence, while six patients developed distant metastases to the liver, lung, or bones without local recurrence. Discussion The relatively high incidence of EGC by Western standards in this study has to be contributed to a very high standard of endoscopy in our region, as well as in our hospital. For the district of Carinthia with about inhabitants, 58 endoscopic units located in 9 hospitals, 3 private medical institutions, 1 social welfare ambulatory, and 45 general gastroenterologic practitioners offer endoscopy for patients with gastrointestinal complaints, which is why 1 endoscopic unit is provided for 9483 inhabitants. Although there are no special screening programs established in Austria, the EGC rate in

5 A. Pertl et al.: Long-term results of early gastric cancer 119 Fig. 2. Five- and 10-year tumor-specific survival probability for 130 patients with EGC: mucosal type, 10 patients (solid line); submucosal type, 60 patients (dashed line) Carinthia is 8.1%, compared to 6.6% in the rest of Austria with a gastric cancer incidence of 41.3 new cases per inhabitants in the year A strong decrease of gastric cancer incidence could be observed in Austria, when one considers the Austrian incidence for the year 1984 to be 82.8/ inhabitants (Austrian head office for statistics). As the percentage of EGC in our surgical institution is 19.2% versus 8.1% in our district, an additional selection factor must be present in regard to the high number of gastric cancer patients operated at our department. In the period , 42% of all documented gastric cancers in our district were concentrated at our surgical department. As no screening of asymptomatic persons is feasible or justified because of the comparably low incidence of gastric cancer in our country, the EGC rates of 40% and greater achieved in Japan cannot be realized. Nevertheless, efforts must be made to assure endoscopy and biopsy for symptomatic patients and those at high risk, for example, patients with a family history of cancer or chronic athrophic gastritis. What is required to detect gastric cancer more often at an early stage is the education of doctors and the public alike in that the onset of any symptoms should be investigated promptly by endoscopy. In our EGC series, nearly 70% of the patients had epigastric complaints in some fashion. The pathohistological findings in our study do not differ from those of larger series in regard to depth of infiltration, incidence of lymph node involvement, or histological grading [28 31]. The distribution in localization is similar to the Japanese series and somewhat different from Western university institutions in which a markedly higher proportion of proximal cancers is present [3 5,9,20,21,23,24]. The fact that lymph node invasion increases drastically with tumor penetration to the submucosa may be explained by the wide network of lymph capillaries found in the submucosa [43 45]. Both the diffuse-type gastric cancer and depth of tumor infiltration did influence the scale of lymph node metastases. Endoscopic ultrasound examination in connection with pathohistological studies in Japan identified criteria for subtypes of mucosal EGC with minimal probability for lymph node metastases [12 17]. This resulted in the introduction of endoscopic therapy modalities for these EGC types of low malignancy, which has been confirmed by long-term results in some specialized Japanese centers [13,15 17]. Although not available for the most of the reported period, endoscopic ultrasound examination in our series would have revealed only a few cases of mucosal EGC for possible endoscopic approach. According to the Japanese criteria for minimally invasive treatment, only 11 cases of our EGC series would have been suitable for possible endoscopic resection. With exception of subtypes of EGC, radical surgery with D2 lymph node dissection, as employed in Japan since the 1950s [3 8,18 24], is of decisive importance for patient outcome. Japanese [3 8,18 24] and European and American studies [27 31,46] as well have demonstrated the worth of lymphadenectomy, not only for EGC. Moreover, immunohistochemical studies for pn0 categories of EGC in conventional paraffin histological examinations could meanwhile establish the value of radical lymph node dissection for EGC, because a high percentage of assessed negative nodes have shown tumor cell microinvolvement, which was recognized as a detrimental independent factor for patient outcome [22 24,47]. Therefore, the phenomenon of stage migration [48] can no longer be the only explanation for the superiority of D2 against D1 lymph node dissection. Overall survival in our series was 91.1% at 5

6 120 A. Pertl et al.: Long-term results of early gastric cancer years and 91.1% at 10 years, 98.1% and 98.1% for mucosal carcinoma and 82.7% and 82.7%, respectively, for submucosal carcinoma. This long-term survival correlates nearly exactly with figures reported by Japanese publications; 5-year survival for mucosal EGC was stated to be between 93% and 96%, whereas survival rates between 82.7% and 87% were reported for submucosal types. The favorable results achieved in our series have to be seen chiefly in context with a consistent radical surgery, employed according to the recommendations of the JRSGC. Furthermore, the familiarity with radical lymph node dissection in an institution with great experience in gastric cancer surgery has contributed to low postoperative morbidity and mortality in comparison to other Western institutions. References 1. Japanese Research Society for Gastric Cancer. The general rules for the gastric cancer study in surgery and pathology. Jpn J Surg 1981;11: Murakami T. Pathological diagnosis. Definition and gross classification of early gastric cancer. Gann Monogr Cancer Res 1971;11: Okuda H, Suzuki S, Suzuki H, Kurokawa K, Kitamura Y, Mitsunaga A, et al. Diagnosis and prognosis of early gastric cancer the experience in Japan. Trop Gastroenterol 1988;9: Maruyama K, Okabayashi K, Kinoshita T. Progress in gastric cancer surgery in Japan and its limits of radicality. World J Surg 1987;11: Maruyama K, Sasako M, Kinoshita T, Okajima K. Effectiveness of systematic lymph node dissection in gastric cancer surgery. In: Nishi M, Ichikawa H, Nakajima T, Maruyama K, Tahara E, editors. Gastric cancer. Berlin Heidelberg New York Tokyo: Springer-Verlag, Kitaoka H, Yoshikawa K, Hirota T, Itabashi M. Surgical treatment of early gastric cancer. Jpn J Clin Oncol 1984;14: Nagayo T. Chronologic development and prognosis of early gastric cancer. In: Nagayo T, editor. Histogenesis and precursors of human gastric cancer: research and practice. Berlin: Springer- Verlag, 1986: Fukotomi H, Sakita T. Analysis of early gastric cancer cases collected from major hospitals and institutes in Japan. Jpn J Clin Oncol 1984;14: Hirota T, Ming Si-Chun, Itabashi M. Pathology in early gastric cancer. In: Nishi M, Ichikawa H, Nakijama T, Maruyama K, Tahara E, editors. Gastric cancer. Berlin Heidelberg New York, Springer-Verlag, Takagi K. Surgery of early gastric cancer in Japan. In: Magenkarzinom, Gall FP, Hermanek P, Hornig O, editors. Bonn: Zuckschwerdt-Verlag, 1986: Kitaoka H, Yoshikawa K, Hirota T, Itabashi M. Surgical treatment of early gastric cancer. Jpn J Clin Oncol 1984;14: Maekawa S, Takeo S, Ikejiri K, Anai H, Saku M. Clinicopathological features of lymph node metastasis in early gastric cancer. Int Surg 1995;80: Hiki Y, Shimao H, Mieno H, Sakakibara Y, Kobayashi N, Saigenji K. Modified treatment of early gastric cancer: evaluation of endoscopic treatment of early gastric cancers with respect to treatment indications groups. World J Surg 1995;19: Yamao T, Shirao K, Ono H, Kondo H, Saito D, Yamaguchi H, et al. Risk factors for lymph node metastasis from intramucosal gastric carcinoma. Cancer (Phila) 1996;77: Takekoshi T, Baba Y, Ota H, Kato Y, Yanagisawa A, Takagi K, et al. Endoscopic resection of early gastric carcinoma: results of a retrospective analysis of 308 cases. Endoscopy 1994;26: Okazaki Y, Tada M. Endoscopic treatment of early gastric cancer. Semin Surg Oncol 1991;7: Tada M, Murakami A, Karita M, Yanai H, Okita K. Endoscopic resection of early gastric cancer. Endoscopy 1993;25: Japanese classification of gastric carcinoma. Japanese Research Society for Gastric Cancer. Nishi M, Omori Y, Miwa K, editors. First English edition. Tokyo: Kanehara; Japan Gastric Cancer Association. Japanese classification of gastric carcinoma. 2 nd edition. Gastric Cancer 1998;1: Inoue K, Tobe T, Kan N, Nio Y, Sakai M, Takeuchi E, Sugiyama T. Problems in the definition and treatment of early gastric cancer. Br J Surg 1991;78: Okamura T, Tsujitami S, Koremaga T, Haraguchi M, Baba H, Hiramoto Y, et al. Lymphadenectomy for cure in patients with early gastric cancer and lymph node metastasis. Am J Surg 1988:155: Baba H, Maehara Y, Okuyama T, Orita H, Anai H, Akazawa K, et al. Lymph node metastasis and macroscopic features in early gastric cancer. Hepatogastroenterology 1994;41: Baba H, Maehara Y, Takeuchi H, Inutsuka S, Okuyama T, Adachi Y, et al. Effect of lymph node dissection on the prognosis in patients with node-negative early gastric cancer. Surgery (St. Louis) 1995;117: Maehara Y, Tomoda M, Tomisaki S, Ohmori M, Baba H, Akazawa K, et al. Surgical treatment and outcome for node negative gastric cancer. Surgery (St. Louis) 1997;121: Hisamichi S, Sugawara N. Mass screening for early gastric cancer by X-ray examination. Jpn J Clin Oncol 1984;14: Maehara Y, Okuyama T, Moriguchi S, Orita H, Kusumoto H, Korenaga D, et al. Prophylactic lymph node dissection in patients with advanced gastric cancer promotes increased survival time. Cancer (Phila) 1992;70: Jatzko G, Lisborg P, Klimpfinger M. Extended lymphadenectomy against early gastric cancer. Jpn J Clin Oncol 1992;22(1): Oliveira FJ, Ferrao H, Furtado E, Batista H, Conceicao L. Early gastric cancer: report of 58 cases. Gastric Cancer 1998;1: Lawrence M, Shiu MH. Early gastric cancer. Twenty-eight year experience. Ann Surg 1991;213: Green PHR, O Toole KM, Slonim D, Wan T, Wej H. Increasing incidence and excellent survival of patients with early gastric cancer. Experience in a United States medical center. Am J Med 1988;85: Sue-Ling HM, Martin I, Griffith J, Ward DC, Quirke P, Johnston D, et al. Early gastric cancer: patients treated in one surgical department. Gut 1992;33: Allum WH, Powell DJ, McConkey CC, Fielding JWL. Gastric cancer: a 25-year review. Br J Surg 1989;76: Bonenkamp JJ, Van de Velde CJH, Songun I, Welvaart K, Sasako M, Hermans J, for the Deutsch Gastric Cancer Group. Randomized comparison of morbidity after D1 and D2 dissection for gastric cancer in 996 Deutsch patients. Lancet 1995;345: Cuschieri A, Fayers P, Fielding J, Craven J, Bancewicz J, Joypaul V, et al. Postoperative morbidity and mortality after D1 and D2 resections for gastric cancer: preliminary results of the MRC randomized controlled surgical trial. Lancet 1996;347: Watanabe H, Jass JR, Sobbin LH. Histological typing of oesophageal and gastric tumors. In: WHO International Histological Classification of Tumors. Second edition. Berlin Heidelberg New York Tokyo: Springer-Verlag, Hermanek P, Sobin LH, editors. UICC TNM classification of malignant tumors. 4th ed. Berlin Heidelberg New York: Springer- Verlag, 1987:39 46.

7 A. Pertl et al.: Long-term results of early gastric cancer International Union against Cancer. In: Sobin LH, Wittekind CH, editors. TNM classification of malignant tumors. 5 th ed. Heidelberg: Springer-Verlag, Wittekind CH, Wagner G, editors. UICC TNM Klassifikation maligner Tumoren. Auflage. Berlin, Heidelberg, New York: Springer-Verlag, Lauren P. The two histological main types of gastric carcinoma: diffuse and so-called intestinal type carcinoma: an attempt at a histo-clinical classification. Acta Pathol Microbiol Scand 1965;64: Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53: Mantel N. Evaluation of survival dates and two new rank order statistics arising in its consideration. Cancer Chemother Rep 1966;50: Cox DR. Regression models and life tables (with discussion). JR Stat Assoc 1972;34: Lehnert T, Erlandson RA, Decosse JJ. Lymph and blood capillaries of the human gastric mucosa. Gastroenterology 1985;89: Listrom MB, Fenoglio-Preiser CM. Lymphatic distribution of the stomach in normal, inflammatory, hyperplastic and neoplastic tissue. Gastroenterology 1987;98: Owen DA. Stomach. In: Sternberg STS, editor. Histology for pathologists. New York: Raven Press, 1992: Jatzko GR, Lisborg PH, Denk H, Klimpfinger M, Stettner H. A 10-year experience with Japanese-type radical lymph node dissection for gastric cancer outside of Japan. Cancer (Phila) 1995; 76: Siewert JR, Kestlmeier R, Busch R, Böttcher K, Roder JD, Müller J, et al. Benefits of D2 lymph node dissection for patients with gastric cancer and pn0 and pn1 lymph node metastases. Br J Surg 1996;83: Bunt AMG, Hermans J, Smit VTHBM, et al. Surgical/pathologicstage migration confounds comparisons of gastric cancer survival rates between Japan and Western countries. J Clin Oncol 1994; 13:19 25.