Prognosis for surgically treated gastric cancer patients is poorer for women than men in all patients under age 50.

From 1965 to 1983, 1031 patients (689 men and 342 women) with advanced gastric cancer underwent gastric resection in our department. A retrospective study was done with special reference to the sex of the patients. The age, tumour size and location, Borrmann type, and histology were considered as the sex-related associations. The survival rate of women under age 50 years was lower than that of men, with a significant difference (P less than 0.01), and the 10-year survival rate was 39.2% for the men and 29.3% for the women. A multivariate analysis showed that the operative curability (relative risk: 2.11), lymph node metastasis (relative risk: 1.37), depth of invasion (relative risk: 1.30) and tumour size (relative risk: 1.05), all significant prognostic factors, differed between the men and women under age 50 years, and the survival rate for women was lower. Thus, early detection of gastric cancer is crucial to improve the survival of women under age 50 years. Postoperative chemotherapy may be considered for those with an advanced gastric cancer.

The emphasis on early diagnosis, operative techniques of extensive lymph node dissection (Kodama et al.. 1981;Boku et al.. 1989) and postoperative chemotherapy has led to a longer survival time for patients with gastric cancer (The Gastrointestinal Tumor Study Group. 1982: Maehara et al.. 1990a. Nevertheless, for a large number of patients with advanced gastric cancer. palliative surgery remains the only choice of treatment and the 5-year survival rate for a stage IV lesion remains around 10% (Korenaga et al.. 1988). Hepatic metastasis. peritoneal dissemination, widespread nodal involvement, depth of invasion and other factors influence the prognosis. but the sex is apparently not a prognostic factor (Baba et al.. 1989: Korenaga et al., 1989: Shiu et al.. 1989. The male to female ratio in patients with gastric cancer is 2:1. while the number of females affected increases in cases of diffuse type carcinomas, and in young patients (Lauren. 1965;Bloss et al.. 1980: Matley et al.. 1988: Mitsudomi et al.. 1989a. Sex hormones play a role in cancers originating in the prostate, breast, or uterine endometrium and also in cancers of the oesophagus and lung (Sugimachi et al.. 1987;Mitsudomi et al.. 1989b). In this context, we analysed various clinical characteristics in patients with gastric cancer following gastric resection, with special reference to the sex of patients.

Patients
This studv was based on a retrospective analysis of 1031 patients with advanced gastric adenocarcinoma and who were treated in the Department of Surgery II, Ky-ushu University Hospital. Fukuoka, Japan. between 1965 and1983. Pathological (Cox. 1972). The level of significance was P<0.05.

Patients
Of the 1031 patients. 25.40o were under age 50 years. 60.6% were in the 51-70 age group and 14.0% were over age 71 years. as shown in Table I. Gastric cancer occurred more frequently in men, a ratio of 2:1 (male female), and in those under age 50 years. the ratio was near 1:1. Table II shows clinicopathological data on all the patients who underwent gastric resection. There were statistical differences in the age. tumour size, location of tumour, Borrmann type and histology between the sexes. We also analysed the clinicopathological factors in three groups. In group 1 patients under age 50 years. statistical differences were noted in tumour size, Borrmann type. histology, depth of invasion. lymph node metastasis and operative curability. In particular. the rate of lymph node metastasis (82.80%o) and serosal invasion (83.6%) was prominent and a non-curative resection was done (50.8%) ( Table III). The cinicopathological factors related to palliative resection are shown in Table IV. The incidence of infiltration at the oral margin: ow( + ) was higher in women under 50 years of age than in men. In the group 2 patients. aged 51-70 years, differences were noted only in tumour size, Borrmann type and histology and in group 3 patients aged over 71 years, there were no apparent differences in clinicopathological factors. To determine which of the many covariates had the most prognostic significance with regard to survival time, a multivariate analysis was made (Maehara et al., 1991b,c). Liver metastasis, operative curability, peritoneal dissemination. operative procedure, lymph node metastasis, depth of invasion and tumour size proved to be independent risk covariates in all patients even those under age 50 years, after gastric resection (Table V).

Survival rates
Postoperative survival curves for all patients are shown in Figure 1. At the time of analysis, the median follow-up time for the 226 patients was 13.5 years. The generalised Wilcoxon test of the two survival patterns revealed no significance. The 10-year survival rate was 34.5% for men and 32.6% for women. In patients under 50 years of age (group 1), the survival rates for the women were lower than those for men, with a statistical difference (P<0.01), as shown in Figure 2. The 10-year survival rate was 39.2% for the men and 29.3% for the women. In group 2 aged 51-70 years and group 3 aged over 71 years. there were no differences in survival rates between the sexes.

Eiscusi
Several clinicopathological factors are involved in determining the prognosis for patients with a gastric cancer (Baba et   Penrtoneal dissemination 3(6.300o) 1(1.60°00 and liver metastasis ow (+) = cancer infiltration at the oral margin; aw (+) = cancer infiltration at the anal margin. Korenaga et al., 1989;Shiu et al., 1989;Maehara et al.. 1991a,c). The sex of the patients was not a significant factor influencing the prognosis; however, clinicopathological factors of gastric cancer between the sexes do differ. The number of women affected was half that of men, but increased 1:1 in the young generation. The undifferentiated type (Sugano et al., 1982) which shows a diffusely infiltrative growth pattern was prominent in women. In young patients and in one case of diffuse types of tumours, the rate of occurrence of gastric cancer increases in the female sex and a high frequency of pregnancy in young women with gastric cancer has been noted (Lauren, 1965;Bloss et al.. 1980;Matley et al., 1988;Mitsudomi et al., 1989a). However, as pregnancy most often occurs in this age group, an associa-  tion could be fortuitous (Matley et al.. 1988). The presence of oestrogen receptors and intracytoplasmic oestradiol in a proportion of patients of all ages fails to explain the preponderance of women among young patients (Nishi et al.. 1987).
Armstrong and Dent (1986) reported that the survival rate of women with gastric cancer exceeds that of men. and Stemmermann and Brown (1974)  ing factors of tumour size. Borrmann type. histology. depth of invasion. Iymph node metastasis and operative curability.
In particular. advanced cases were dominant in women and most often a non-curative resection was done. Therefore. the statistical lower survival rate for the women under age 50 years is expected to be influenced by operative curability. lymph node metastasis. depth of invasion and tumour size. all significant prognostic factors (Table IV). Clinical diagnosis mav be made late and the young age of the women is the major deterrent to early diagnosis (Bloss et al.. 1980). As tumour cells left behind at surgery may proliferate rapidly in non-curatively resected cases (Schabel. 1975: Gunduz et al.. 1979. the potential to control the remaining tumour foci is significantly reduced by postponing chemotherapy to the later postoperative period (Tubiana & Malaise. 1976: Douglass. 1985. We found that the undifferentiated gastric cancer tissue is more sensitive to anti-tumour drugs than is differentiated cancer tissue. in *-itro (Maehara et al.. 1987) and the undifferentiated type is dominant in women under age 50 years. Therefore. postoperative chemotherapy. for example with the combination of mitomycin C. fluorinated pyrimidine and the immunomodulator PSK. should be initiated in a few days and continued for 1 year in the postoperative period (Maehara et al.. 1990a.b).
We thank M. Ohara for comments.