Special Feature

Gastric Carcinoma: Changing Trends in Therapy

By Arden Morris, MD, and John T. Vetto, MD

The incidence of gastric cancer has declined four-fold in the United States since World War II, yet it remains the eighth leading cause of cancer-related death in this country and has an overall, post-operative, five-year survival rate of only 20%. In contrast, in Japan, where gastric cancer is the leading cause of cancer-related death (with an incidence 5 times higher than the United States), post-surgical cure rates approximate 50%. These substantially different surgical cure rates have been the subject of much speculation and certainly merit further investigation.

Compared to Western treatment strategies, the Japanese approach to the therapy of gastric cancer includes the widespread use of endoscopic screening (yearly after age 50) and aggressive surgical resection of the tumor with an extensive en bloc lymph node dissection, techniques that have been employed in Japan since the 1960s. At present, uncertainty remains regarding the following issues: a) Is there genetic or geographic variability in the biological behavior of the disease?; b) What is the appropriate extent of surgical resection of the stomach, lymph nodes, and adjacent organs?; and c) What is the efficacy of adjuvant therapy?

Biological behavior of gastric cancer: Two different diseases?

During the last 50 years, the most common location for gastric carcinoma in the stomach appears to have migrated proximally. This change carries implications for both prognosis and the type of surgical resection needed. Currently, 40% of gastric carcinomas in the United States are found in the proximal stomach. Only 15% of carcinomas were similarly located during the 1960s. Partially due to its diffuse histology (implying widespread intramural invasion at the time of diagnosis), proximal gastric cancer is associated with a dismal 10% five-year postoperative survival rate in Western countries.1 In contrast, distal gastric cancer usually involves less aggressive intestinal-type pathology, is associated with an older mean age at onset, and is more common in the Japanese population. Some investigators, citing the differences in location, histology, and prognosis, believe that proximal and distal gastric carcinomas may represent two different diseases, thus providing a possible explanation for the differences observed in Western and Japanese outcomes. This seems an unlikely explanation because in their comparison of 1396 Japanese gastric cancer patients diagnosed between 1980 and 1984 and 1036 patients diagnosed between 1961 and 1965, Kampschoer and colleagues found the same trends in tumor histology and localization that were observed in Western patients.2 Interestingly, the overall reduction in the incidence of gastric cancer in the United States appears to be due solely to fewer cases of distal gastric cancer.3 Although the specific reasons for the decreased incidence and shifts in the location of primary gastric cancers remain unknown, the most often-quoted hypotheses implicate H. pylori, which is linked to atrophic gastritis and dietary changes, which are supported by migration statistics.

Extent of resection

Due to the rich capillary and lymphatic network of the stomach wall, extensive intramural spread has often already occurred at the time of gastric cancer diagnosis. For this reason, standard surgical protocol mandates a total gastrectomy for proximal tumors, which tend to have increased intramural spread at presentation, and an 80% gastrectomy for distal tumors, with no less than a 5 cm disease-free margin in order to ensure that all tumor is removed. Surgery remains the most effective treatment for gastric cancer, and residual disease following resection is a significant negative prognostic factor.1, 3-6

Considerable controversy exists regarding the appropriate extent of lymphadenectomy, mainly due to disagreements as to whether resection of involved lymph nodes is of only prognostic value or also has therapeutic potential. This is an important question because even early gastric cancer, in which no serosal penetration has occurred, exhibits a 15% incidence of lymph node metastasis at diagnosis. Japanese authors cite impressive overall survival rates to support their belief that lymphadenectomy is an important therapeutic procedure. In Japan, surgeons routinely perform an extended lymphadenectomy (D2), which removes not only perigastric nodes (N1), but also celiac, peripyloric, peripancreatic, perisplenic, and retroperitoneal nodes (N2, N3), with a reportedly low operative mortality rate (approximately 3%).7

Enthusiasm for extended lymphadenectomy is echoed by some Western investigators. A prospective, multicenter trial by the German Gastric Carcinoma Study Group followed 1654 gastric carcinoma patients after curative gastric resection.4,5 In this trial, operative techniques were standardized by consensus meetings and teaching videos. One thousand ninety-six patients underwent a radical lymph node dissection (defined as ³ 26 nodes in the specimen) in addition to gastrectomy. This group did not have an increase in morbidity or mortality compared to 558 patients subjected to "standard" lymph node dissection (< 26 nodes) with gastrectomy. Multivariate analysis revealed that extended lymphadenectomy was the most important predictor of survival and that radical dissection provided a significant survival benefit for patients with stage II or IIIa disease.4 Roder and others tempered this conclusion slightly, noting that patients with stage II or IIIa tumors comprised only 30% of the total group, and, therefore, the overall effect of radical lymph node dissection could not be determined.5 Roder did note that, in this trial, the ratio of involved to resected nodes was important and associated with a statistically significant deterioration of the prognosis when more than 20% of the lymph nodes resected were involved.

In the United States, Volpe and collaborators published a retrospective review comparing 101 patients who underwent either a limited (D1, which includes only N1 nodes) or an extended node dissection (D2, which includes N1 and N2 nodes) for T2-4, N0-1 gastric cancer.8 Although an overall operative mortality for the entire study group of 5% was reported, a comparison of the morbidity and mortality between the two groups was notably absent. Based on their five-year survival rates of 16% and 44%, following limited or extended dissections, respectively (P = 0.05), Volpe et al endorsed the more extensive resection.

In contrast to the broad recommendations of Volpe and others, a large, prospective, observational study conducted by the German Stomach Cancer TNM Study Group found a non-statistically significant survival benefit with "systematic" lymphadenectomy (D2) for stage I patients only.9 Also in contrast to the findings of Volpe et al, Wanebo and colleagues reviewed prospectively gathered data from 3804 non-randomized, curatively resected patients and found that respective five-year survival rates were 35.6%, 30%, and 26.3% after gastric resection plus D0, D1, and D2 lymphadenectomies.10 Subgroup analysis also showed no improvement in survival for patients who had an extensive (D2) lymphadenectomy.

Critics of the therapeutic value of routine extended node dissection point to several questions incompletely addressed by D2 enthusiasts. The most commonly raised issue concerns "stage migration" (sometimes referred to as the "Will Rogers phenomenon" based upon his observation that "when the Okies left Oklahoma and moved to California, they raised the IQ of both states"). This concept refers to the improvements in staging accuracy that lead to the upstaging of patients and to the an impression of better stage-per-stage outcome. In a randomized, prospective trial of 473 patients undergoing D1 or D2 curative resection, Bunt and associates discovered 30% upstaging to N2 status of only patients who underwent D2.11 Additional confounding factors in studies of D2 lymphadenectomy include selective extension of lymphadenectomy outside the designated area of clearance ("contamination"), incomplete lymphadenectomy at designated areas ("noncompliance"), and examination of more than the specified number of nodes at a designated level ("diligence").

In an effort to compare disease-free interval, survival benefit, and morbidity and mortality of D1 and D2 operations in a valid fashion, two large phase III studies (the Dutch Gastric Cancer Trial and the Medical Research Council [MRC] Gastric Cancer Trial) with rigorously planned surgical quality control have been conducted. A recent update of the 996 patients with stages I-III gastric cancer in the Dutch trial revealed markedly higher morbidity and operative mortality rates of 43% and 10%, respectively, for patients undergoing gastric resection with D2 lymphadenectomy as compared with 25% and 4% for those undergoing D1 lymphadenectomy (P = 0.004), despite earlier reports of no differences between groups.12-14 The MRC trial was designed to address these same issues for patients with stage I and II disease. Although long-term survival data are not yet available, morbidity and mortality data are consistent with those from the Dutch trial.15 Post operative morbidity and mortality were significantly worse for the D2 group vs. the D1 group, with rates of 46% vs. 28% and 13% vs. 6.5%, respectively. Postoperative morbidity and mortality were associated more with pancreatectomy and splenectomy, rather than with the extended lymph node dissection specifically.

Adjuvant therapy

Fifty percent of patients undergoing resection for cure will recur and die of metastatic disease, presumably due to residual micrometastases. Furthermore, approximately 50% of patients with gastric cancer have advanced disease and are inoperable at the time of diagnosis. Accordingly, despite scant evidence for efficacy, interest in adjuvant therapy of gastric cancer remains high.

Single agent chemotherapy has been associated with an objective response (i.e., > 50% reduction in tumor size) in approximately 15% of patients with advanced disease.3 Notably, 5-FU reportedly provides a response rate of up to 20%. Combination chemotherapy typically produces better results; for example, a 30-40% overall response rate has been reported with use of combination 5-FU, doxorubicin, and mitomycin C; even better response rates have been reported when high-dose methotrexate is substituted for mitomycin C.

Unfortunately, caveats to these data abound. In 1993, Hermans and colleagues conducted a formal meta-analysis of 11 adjuvant treatment studies published since 1980 containing a surgery-alone control arm compared to a surgery plus chemotherapy arm.16 Although adjuvant chemotherapy offered some utility in phase II trials, no overall survival benefit could be shown, and it is not employed routinely for patients in the United States outside of a clinical trial. The differences in gastric cancer in Japan and the West are not solely in the area of surgery. For example, Japanese trials claim survival advantages for adjuvant therapy recipients, treated with either mitomycin C or mitomycin C, 5-FU, and cytosine arabinoside. These studies have been criticized for inclusion of many early stage patients and lack of prospectively identified control groups.1,3 However, adjuvant chemotherapy would be considered standard therapy in Japan.

Treatment with adjuvant radiation therapy alone for gastric cancer provides neither additional control of local disease nor survival benefit.1,3 However, concurrent use of 5-FU as a radiation-sensitizing agent increased the post operative five-year survival rate in patients with advanced local disease from 6% to 18% in a phase III Gastrointestinal Tumor Study Group trial. In light of these and other relatively encouraging results, a U.S. intergroup trial (INT 0116) has begun accruing patients with resectable stage II-IV gastric carcinoma to a two-arm trial of either surgery alone or surgery with post operative 5-FU, leucovorin, and radiation therapy.1 Finally, initial results of small phase II neoadjuvant therapy trials at the NCI show some promise of efficacy and remain to be validated in phase III trials.16


1. While proximal and distal gastric cancer are associated with different histology, behavior, prognosis, and geographic incidence, and may indeed be different diseases, gastric cancer in both Western and Japanese patients is exhibiting the same trend toward more proximal tumors.

2. For staging purposes alone, an "as needed" lymphadenectomy (removing only involved lymph nodes as determined by palpation) is insufficient since macroscopically evaluated nodes are classified incorrectly in 7-23% of patients. A D2 lymphadenectomy clearly improves staging accuracy but is associated with higher morbidity and mortality, as shown in both the Dutch and MRC trials.

3. The potential therapeutic value of a D2 lymphadenectomy remains unproven, pending survival outcomes from the Dutch and MRC trials, the only western phase III trials examining the effect this operation has on the outcome of surgically treated gastric cancer.

4. Accordingly, while a D2 lymphadenectomy is the standard of treatment for Japanese patients, evidence accumulated thus far does not support routine replacement of D1 by D2 lymphadenectomy for Western patients.

5. Adjuvant therapy should be limited to prospective studies, although this may change pending results of a recent combined modality phase III trial (INT 0116). (Dr. Morris is a Resident, Department of Surgery, Oregon Health Sciences University, and Dr. Vetto is Assistant Professor of Surgery, Section of Surgical Oncology, Oregon Health Sciences University, Portland.)


1. Macdonald JS, Schnall SF. World J Surg 1995;19: 221-225.

2. Kampschoer GHM, et al. Br J Surg 1989;76:914.

3. Fuchs CS, Mayer RJ. N Engl J Med 1995;333:32-41.

4. Siewert JR, et al. Br J Surg 1993;80:1015-1018.

5. Roder JD, et al. Cancer 1993;72:2089-2097.

6. Soreide JA, et al. Arch Surg 1996;131:481-487.

7. Noguchi M, Miyazaki I. Br J Surg 1996;83:156-161.

8. Volpe CM, et al. J Am Coll Surg 1995;181:56-64.

9. Keller E, et al. J Am Coll Surg 1994;178:38-46.

10. Wanebo HJ, et al. J Am Coll Surg 1996;183:616-624.

11. Bunt AMG, et al. J Clin Oncol 1995;13:19-25.

12. Bonenkamp JJ, et al. Eur J Surg 1992; 158:413-418.

13. Roder JD, et al. World J Surg 1995;19:546-553.

14. Bonenkamp JJ, et al. Lancet 1995;345:745-748.

15. Cuschieri A, et al. Lancet 1996;347:995-999.

16. Hermans J, et al. J Clin Oncol 1993;11:1441-1447.