Abstract
In 351 patients with gastric carcinomas resected for cure, the relationship between macroscopic and microscopic features and survival was studied by univariate and multivariate analyses. In the multivariate survival analysis with covariates according to the Cox regression model, in early cancer all significant correlations to survival rate are covered by the stage grouping according to the UICC schedule of 1987. In advanced gastric carcinoma the UICC stage and, in addition, the Borrmann type and the intensity of cellular infiltration are effective. By additional consideration of these two variables an extended pathological staging schedule is proposed. It has the advantage of better discrimination between patients who differ in prognosis and seems to improve the prognostic prediction of outcome. Testing of this extended staging system in larger collectives is recommended.