602 Risk Factors for Lymph Node Metastasis from lntramucosal Gastric Carcinoma Takekazu Yamao, M.D.' Kuniaki Shirao, M.o.' Hiroyuki Ono, M.D.' Hitoshi Kondo, M.D? Daizo Saito, M.D? Hajime Yamaguchi, M.D.* Mitsuru Sasako, M.D? Takeshi Sano, M . D . ~ Atsushi Ochiai, M.D.3 Shigeaki Yoshida, M.D? ' Department of Internal Medicine, National Cancer Center Hospital, Tokyo, Japan. * Department of Surgery, National Cancer Center Hospital, Tokyo, Japan. Department of Pathology, National Cancer Center Hospital, Tokyo, Japan. Department of Internal Medicine, National Cancer Center Hospital East, Chiba. Japan. BACKGROUND. Although regional lymph node metastasis from intramucosal early gastric carcinoma (EGC) is rare, it is very important to clarify the characteristics of patients having lyniph nodal metastases in order to determine appropriate therapy. METHODS. The authors investigated 1196 patients with solitary intramucosal EGC who underwent resection at the National Cancer Center Hospital in Tokyo, with special reference to lymph node metastases. Eight clinicopathologic factors (age, sex, tumor: size, location, macroscopic type, histologic type, histologic ulceration of the tumor, and lymphatic vessel invasion) were investigated by univariate and multivariate analyses for their possible relationship to lymph node metastasis. RESULTS. Lymph node metastases were found in 43 patients (3.5%). Univariate analysis revealed that younger age (< 5 7 years), macroscopic depressed type, larger tumor size ( 2 30 mm), undifferentiated histologic type, histologic ulceration of the carcinoma, and lymphatic vessel invasion had a significant association with regional lyniph node metastasis. Multivariate analysis revealed that lymphatic vessel invasion, histologic ulceration of the tumor, and larger size ( 2 3 0 mm) were independent risk factors for regional lymph node metastasis. The incidence of lymph node metastasis from intramucosal EGC negative for these 3 risk factors was only 0.36% (1 in 277 patients). CONCLUSIONS. Lymphadenectomy is unnecessary for patients with small intramucosal EGC with neither histologic ulceration of the tumor nor lymphatic vessel invasion because the incidence of regional lymph node metastasis is extremely low in those patients. The therapeutic options for such patients would be local resection or endoscopic resection. Cancer 1996; 77502-6. 0 1996 American Cancer Society. KEYWORDS: early gastric carcinoma, regional lymph node metastasis, multivariate analysis, risk factor, lymphadenectomy, endoscopic treatment. E The work was supported in part by a Grant-inAid for Cancer Research (5-18 and 7-34) from Ministry of Health and Welfare, Japan. Address for reprints: Takekazu Yamao, M.D., Department of Internal Medicine, National Cancer Center Hospital, 5-1-1 Tsukiji, Chuo-Ku, Tokyo 104, Japan. Received April 20, 1995; revision received June 27, 1995; accepted June 27, 1995. 0 1996 American Cancer Society arly gastric carcinoma (EGC)is defined as that confined to the mucosa or submucosa regardless of the presence or absence of regional lymph node metastasis.' As a result of diagnostic development, including X-ray and endoscopy, the detection of EGC has been increasing in Japan. The prognosis of patients with EGC has improved with surgical treatment.' Gastrectomy with complete removal of primary and secondary lymph nodes has been the standard operation for EGC in Japan."' This surgical strategy provides an excellent therapeutic outcome; the 5-year survival rate after curative gastrectomy is more than 90% in Japan.',"' Although one of the most important factors for the prognosis of patients with EGC is the presence or absence of regional lymph node metastasis,"-" the incidence of metastasis of intramucosal EGC is approximately 3% and is 20% in submucosal EGC.l3-Ifi Thus, it should be possible to modify the therapeutic strategy for intramucosal EGC, taking into consideration the Nodal Metastasis of Early Gastric CarcinomaNarnao et al. risk of surgery and quality of life for the patients. The purpose of this study is to clarify independent risk factors for regional lymph node metastasis from intramucosal EGC that would be useful for therapeutic determination. 603 TABLE I Relationship between Clinicopathologic Factors and Regional Lymph Node Metastasis from Intramucosal EGC, and Univariate Analysis Results Regional lymph node metastases MATERIALS AND METHODS Patients and Specimens A total of 2265 patients with EGC underwent gastrectomy at the National Cancer Center Hospital in Tokyo between 1963 and 1993. The resected stomachs were opened along the greater or lesser curvatures, pinned on a plate, and fixed in formalin. The specimens containing the tumors and the surrounding gastric wall were cut into multiple slices, principally parallel to the lesser curvature, at an interval of 5 mm. Dissected lymph nodes were also fixed in formalin, cut into two, and embedded in paraffin. Each slice embedded in paraffin was prepared for histologic examination. All microscopic sections were stained with hematoxylin and eosin. All histologic examinations were made by pathologists. Of the 1196 specimens proven to be solitary intramucosal cancer, 43 (3.5%) revealed regional lymph node metastasis histologically. The distribution of positive lymph nodes in all 43 specimens was to primary lymph nodes ( n l ) in 31 (72%), secondary lymph nodes (1121 in 7 (16%), and unknown in 5 (12%),according to the Japanese Classification of Gastric Cancer outlined by the Japanese Research Society for Gastric Cancer.” Eight clinicopathologic factors were investigated by means of univariate and multivariate analyses for their possible association with lymph node metastasis. Each factor was divided into 2 or 3 subgroups: age younger than 57 years or 57 years or older (the mean age of the patients was 57 years); sex; location of the primary lesion as either C (upper third of stomach), M (middle third of stomach), or A (lower third of stomach), according to the Japanese Classification; l 7 macroscopic type (protruded type or depressed type); tumor size smaller than 30 mm or 30 mm or larger in diameter (the mean diameter of the tumor was 30 mm in diarneter); histologic type (differentiated type or undifferentiated type); histologic ulceration of the tumor (negative 0 1 positive); and lymphatic vessel invasion (negative or positive). The primary lesions were classified macroscopically according to the Japanese Classification” as follows: Type I (protruded type), IIa (superficial elevated type), 1Ib (flat type), IIc (superficial depressed type), and Ill (excavated type). In the current analysis, we defined macroscopic protruded type as Type I, Type Ha, or combined type with Type I or Type Ha, such as Ha plus IIc type, and defined macroscopic depressed type as Type IIb, IIc, 111, or combined type with Type IIb, IIc, or 111, such as IIc plus 1Ib or IIc plus 111. As for the histologic classification, tubular adenocarcinoma and papillary adenocarcinoma were classified as differentiated type, and poorly differen- Clinicopathologicfactor Negative Positive Percentage “of positive” P value Age (yr) 4 7 257 Sex Male Female Location in stomach C (upper third) M (middle third] A (lower third) Macroscopic type Protruded Depressed Unknown* 554 599 30 13 5.1% 2.1% 0.005 754 399 25 18 3.2% 3.3% 0.327 103 683 367 ) 1.970 3.5% 0.536 25 16 4.201 7 0.8% 4.3% <n.ooi 41 620 517 16 12 31 l.9% 5.7% <o.nni 667 486 12 31 1.8% 6.0% m n i 416 i2l 16 2 31 0.48% 5.4% <n.ooi 1036 30 3 235 912 6 Size (inin) <30 230 Unknown’ Histological type Differentiated Undifferentiated Histological ulceration Negative Positive Unknown* Lymphatic vessel invasion Negative Positive Unknown* 6 3.7% 33.3% <0.001 111 EGC early gastric carcinoma. ’ Unknown cases were excluded from statistical analysis tiated adenocarcinoma, mucinous adenocarcinoma, and signet ring cell carcinoma were classified as undifferentiated type. Statistical Analysis The SAS program (SAS Institute, Inc., SAS Campus Drive, Cary, NC) was used for all analyses. The association of lymph node metastasis with clinicopathologic variables was assessed using a simple chi-square test. Multivariate analysis was carried out using a logistic model with a stepwise method. A level of P < 0.05 was taken as significant. RESULTS Univariate Analysis of Risk Factors for Regional Lymph Node Metastasis from lntramucosal EGC The associations between clinicopathologic factors and the presence or absence of lymph node metastasis are 604 CANCER February 15, 1996 / Volume 77 / Number 4 TABLE 2 Independent Risk Factors for Regional Lymph Node Metastasis from lntramucosal EGC Clinicopathologicfactor (observed value) Lymphatic vessel invasion (negative or positive) Histologic ulceration (negative or positive) Size of lesion (c30 nini or 2 30 mml Relative risk (95%C.I.*) S.E.** P value 14.33 (2.98-68.841 0.8007 0.0009 6.09 (1.38-26.821 0.7564 0.0169 3.11 (1.4-7.57) 0.4177 0.0038 tCL: rarl, gasirir c~ircit~i)n~~i: (:I: ciinfidcnce i n l e n d SE: standard rrriit summarized in Table 1. Age younger than 57 years, macroscopic depressed type, tumor size larger than 30 m m in diameter, undifferentiated histologic type, histologic ulceration of the tumor, and lymphatic vessel invasion had a significant association with lymph node metastasis by chi-square test. Multivariate Analysis of Risk Factors for Regional Lymph Node Metastasis from lntramucosal EGC We chose six clinicopathologic factors (age, macroscopic type, tumor size, histologic type, histologic ulceration of the tumor, and lymphatic vessel invasion) that had a significant association for lymph node metastasis in the univariate analysis described above. As shown in Table 2, the independent risk factors for regional lymph node nietastases were found to be lymphatic vessel invasion, a tumor size larger than 30 nim in diameter, and histologic ulceration of the tumor. The incidence of regional lymph node metastases of intramucosal EGC specimens that were negative for these 3 risk factors was only 0.36% (one of 277 specimens). DISCUSSION The presence or absence of regional lymph node nietastasis is one of the most important factors for the prognosis of patients with EGC.""' Lymph node metastasis of intramucosal EGC is rare. I n general, lymph node metastasis of intramucosal EGC has been thought to be an exceptional event. Although the number of resected EGCs has increased in Japan, the clinicopathologic characteristics of intramucosal EGC with regional lymph node metastasis have been unclear because of their low incidence. Although gastrectoniy with lymph node dissection has been the standard treatment for EGC in Japan,"-bpatients with EGC who are free of regional lymph node metastasis can be cured without lymph node dissection. In addition, with the recent developinent of endoscopic mucosal resection for EGC, patients with n o lymph nodal metastasis can be treated using endoscopy without increasing the risk of recurrence."' Thus, considering the low incidence of lymph node metastasis, the risk of surgery, and quality of life for the patients, there is much to be said for modifying the therapeutic strategy for the cure of intramucosal EGC on the basis of patients' risk factors for regional lymph node metastasis. This is the first report with reference to risk factors for lymph node metastasis of intramucosal EGC by multivariate analysis of a large number of patients in a single institution. In this study, regional lymph node metastases were found in 3.5% of intramucosal carcinomas, which was not much different from previous reports.''-'b Sano et al. reported 14 cases (3.3%)of intramucosal gastric carcinomas with lymph node meta~tases.'~ They noted that 12 of the 14 lesions were of the macroscopically depressed type and were accompanied by histologic ulceration of the tumor and that one was a large (55 m m in diameter) protruded type. Furthermore, no difference was found in the distribution of histologic types. Korenaga et al. reported that lymph node metastasis was found in 11 of 568 cases (1.9%) of intramucosal EGC and that all lesions were larger than 20 mm in diameter." Nine lesions were of the undifferentiated type and six were accompanied by histologic ulceration of the tumor. In the present study, similar findings were observed in the characteristics of intramucosal carcinoma with lymph node metastasis. Lymph node metastases were significantly associated with a younger age, macroscopically depressed lesion, larger tumor size, undifferentiated histologic type, histologic ulceration of the tumor, and lymphatic vessel invasion in the iinivariate analysis. Furthermore, multivariate analysis revealed that lymphatic vessel invasion, histologic ulceration of the tumor, and larger tumor size were independent risk factors for lymph node metastases of intramucosal EGC. It is readily understandable that lymphatic vessel invasion is one of the risk factors for lymph node metastases because its presence indicates that cancer cells have already permeated to lymph flow. Lymphatic vessels are few in the mucosal layer, whereas the submucosal layer is rich in them. The existence of histologic ulceration within the tumor indicates the destruction of the muscularis mucosa that acts as a barrier against the lymphatic vessel invasion. The speculation proposed by Sane'" makes it plausible that the destruction of the muscularis mucosa by histologic ulceration could cause a n interchange between the lymph flow in the mucosa and submucosa, which could result in a n increase in the risk of regional lymph node metastasis. Additionally, the breakdown of the muscularis mucosa also means the possibility that a preceding invasion to the submucosal layer has already occurred. The results of multivariate analyses of risk factors provide useful information in the clinical decision-mak- Nodal Metastasis of Early Gastric Carcinornaflarnao et al. ing processes. In the present study, lymphatic vessel invasion, histologic ulceration of the tumor, and larger tumor diameter were the independent risk factors for regional lymph node metastasis from intramucosal EGC. The present findings theoretically provide useful information for the determination of how extensively lymphadenectomy should be performed in individual patients with intramucosal EGC. The incidence of regional lymph node metastases in intramucosal EGC that was negative for lymphatic vessel invasion, histologic ulceration of the tumor, and a tumor less than 30 m m in diameter was 3.8% (40 metastases of 1076 tumors), 0.48% (2 metastases of 418 tumors) and 1.9% (12 metastases of 632 tumors) respectively (‘Table 1). The incidence of lymph node metastases of intramucosal EGC in which tumor size was less than 30 m m in diameter with neither histologic ulceration of the tumor nor lymphatic vessel invasion was only 0.36% (1 of 277). Thus, the conventional view that gastrectomy with extensive lymphadenectomy should be the standard surgical procedure for every EGC may be thrown into question because the incidence of regional lymph node metastasis is extremely low in patients with intramucosal EGC who are negative for those risk factors. Even without Iymphadenectomy, patients without those risk factors can be highly curable and may be good candidates for local resection, preserving the function of the cardia and the pylorus. ‘Those same patients also may be good candidates for endoscopic resection, especially if they have surgical risks such as cardiovascular disorders or renal fiilure. ‘Ihe clinical issue is the selective performance of surgery without lymphadenectomy or endoscopic resection for patients with EGC who are free of lymph node metastasis. The appropriate selection of patients applicable to the aforementioned therapeutic strategies depends o n how accurately diagnosis can be made for depth of invasion, tumor size, histologic ulceration of the tumor, and lymphatic vessel invasion. The size of the primary lesion can be determined by routine endoscopy and the histologic ulceration of the tumor can be estimated by routine endoscopy and endoscopic ultrasonography by detecting any sign of converging folds or submucosal Conversely, lymphatic vessel invasion and minute submucosal invasion cannot be diagnosed without histologic examination of the primary tumor. However, endoscopic “strip biopsy” or endoscopic mucosal resection (EMR), introduced by Tada et al.,L’should provide more accurate information on the primary lesion because histologic evaluations are possible in the specimens resected endoscopically. Because the specimens resected by EMR include the submucosal layer, we can evaluate by histologic examination not only the depth of tumor invasion but also lymphatic vessel invasion and histologic ulceration of the tumor. Therefore, by combining these clinical data, 605 it should be possible to select patients with EGC who are likely to be free of regional lymph node metastasis. Furthermore, EMR can be not only the diagnostic procedure described above but also a radical treatment for selected patients with EGC who are negative for these risk factors. When the primary lesion is completely removed endoscopically and the resected specimen is histalogically revealed to be free of these risk factors, EMR can be a curative treatment. It appears that EMR is an important diagnostic procedure for therapeutic determination of small EGC and could be a first choice of treatment for patients with surgical risk. We conclude that the independent risk factors for regional lymph node metastasis of intramucosal EGC are larger tumor size, histologic ulceration of the tumor, and lymphatic vessel invasion, and that because of the extremely low incidence of lymph node metastasis, a lymphadenectomy is unnecessary for patients with EGC who are negative for these factors. The therapeutic options for such patients would be local resection or endoscopic resection. REFERENCES 1. Japanese Research Society for Gastric Cancer. The general rules for gastric cancer study in surgeiy an d pathology. lpti I Surg 1981; 11:127-39. 2. ltoh H, Oohata Y, Nakainura K, Nagata T, Mibu R, Nakayania F. Coniplete ten-year postgastrectoniy follow-up of early gastric cancer. A m 1 Siirg 1989; 158:14-6. 3. Okarnura T, Tsujitani S, Korenaga D, Haraguchi M. Baba H, Hirarnoto Y, et al. Lymphadenectomy for cure in patients with early gastric cancer and lymph node metastasis. A i n 1 Siirg 1988; 155:476-80. 4. Maruyama K, Okabayashi K, Kinoshita T. Progress in gastric cancer surgery in Japan and its limits of radicalit)!. Wor-/dJ Surg 1987; 11:418-25. 5. Noguchi Y, lmada T, Matsuinoto A, Coit DG, Brennan MF. Radical surgery for gastric cancer: a review of the Japanese experience. Cancer 1989;64:2053-62. 6 . Ohta H, Nogiichi Y, Takagi K, Nishi M, Kajitani T, Kato Y. Early gastric carcinoma with special reference to niacroscopic classification. Cancer 1987;60:1099- 110G. 7. K i t 0 T, Yamamura Y, Kobayashi S. Surgical treatment of early gastric cancer. Anticancer Res 1989;8:335-8. 8. Koga S, Kaibara N, Tamura H, Nishidoi H, Kirnura 0. Cause of late postoperative death in patients with early gastric cancer: special reference to recurrence and the incidence of metachronous primary cancer in other organ. Siirgety 1984;6:511-6. 9. Kitaoka T, Yoshikawa K, Hirota T. Itabashi M. Surgical treatment of early gastric cancer. J p n J C h i Oncol 1984;14:28393. 10. Kidokoro T. Frequency of resection, metastasis and five-year survival rate of early gastric carciiionia in a surgical clinic. In: Murakami T, editor. GANN monograph on cancer research 11. Early gastric cancer. Tokyo: IJniversity of Tokyo Press. 1971:45-9. 606 CANCER February 15, 1996 / Volume 77 / Number 4 11. Shiu MH, Moore E, Sanders M, Huvos A, Freedman B, Goodbold J, et al. Influence of extent of resection on survival after curative treatment of gastric carcinoma. Arch Surg 1987;122:1347-51. 12. Sano T, Sasako M, Kinoshita T, Maruyama K. Recurrence of early gastric cancer - Follow-up of 1475 patients and review of Japanese literature. Cmcer 1993;72:3174-8. 13. Fukutomi H, Sakita T. Analysis of early gastric cancer cases collected from major hospitals and institutes in Japan. Jpn J CIirl OTLCO~ 1984;14:169-79. 14. Sano T, Kobori 0. Muto T. Lymph node metastasis from early gastric cancer: endoscopic resection of tumour. Br J SUrg 1992;79:241-4. 15. Korenaga D, Haraguchi M, Tsujitani S, Okamura T, Tamada R, Sugimachi K. Clinicopathological features of mucosal carcinoma of the stomach with lymph node metastasis in eleven patients. Br J Sirig 1986;73:431-3. 16. Ichiyoshi Y, Toda ‘r, Minamisono Y, Nagasaki S, Yakeishi Y, Sugimachi K. Recurrence in early gastric cancer. Surgery 1990;3:489-95. 17. Japanese Research Society for Gastric Cancer. Japanese Classification of Gastric Cancer. Tokyo: Kanehara, 1995. 18. Fukase K, Matsuda T, Suzuki M, Toda H, Okuyama Y, Sakai J , et al. Evaluation of the efficacy of endoscopic treatment for early gastric cancer considered in terms of long-term prognosis - A comparison with surgical treatment. Dig Endose 1994;6:241-7. 19. Hamura K, Sumii K, Inoue K, Teshima H, Kajiyama G. Endoscopic therapy in patients with inoperable early gastric cancer. A m J Gastroenterol 1990;85:522-6. 20. Sano R. Pathological analysis of 300 cases of early gastric cancer. With special reference to cancer associated with ulcers. In: Murakami T, editor. Gann monograph on cancer research 11. Early gastric cancer. Tokyo: University of Tokyo Press, 1971:81-9. 21. Akahoshi K, Misawa T, Fujishima H, Chijiiwa Y, Maruoka A, Nawata H. Preoperative evaluation of gastric cancer by endoscopic ultrasound. Gut 1991;32:479-82. 22. Maruta S, Tsukamoto Y, Niwa Y, Goto H, Hase S, Yoshikane H. Endoscopic ultrasonography for assessing the horizontal extent of invasive gastric cancinoma. Am J Gastroenferol 1993;88:555-9. 23. Tada M, Shimada M, Murakami F. Development of the stripoff biopsy. Gastroenferol Endosc 1984;26:833-9.