Morphological predictors of survival in advanced gastric carcinoma univariate and multivariate analysis.

In 76 curative resected advanced gastric carcinomas, the relationships between macro- and microscopy of the carcinoma and the survival rate were studied by univariate and multivariate survival analyses. In the Kaplan-Meier survival analysis (product limit estimator), significant influence on survival rate was found for tumor size, Laurén type, number of lymphocytes, tumor fibrosis, and N stage. In the multivariate survival analysis of covariates according to the Cox regression model, in macroscopically evaluable variables the tumor size was effective, in bioptically evaluable variables the number of plasma cells and the histological type. Concerning the primary tumor, the same set of variables presents significant correlations to survival time. Adding the involvement of lymph nodes, Laurén type, and N stage express all significant correlations of the tumor to survival rate. In parametric multivariate stepwise regression analyses of survival time, all variables working in the Cox proportional hazard model were very ineffective. But adaption of the Kaplan-Meier test to the effective variables of multivariate survival analysis elucidates the tight relationships between survival rate and these variables. No distinct relationships are present between variables of primary tumor and presence of lymph nodes metastases or N stage respectively.