[Prognostic relevance of biological and molecular markers in oncology. Criteria for planning and interpreting studies].

Experts on different specialties (surgery, pathology, molecular biology and medical statistics) discussed the evaluation of prognostic factors during a workshop. The most important results presented are: (1) the prognostic relevance of new biological or molecular markers must be evaluated in a phase-III prognostic study. To establish such a marker in the UICC-TNM-classification in at least two different centers, two multivariate analyses according to defined criteria are necessary; (2) the standards of laboratory methods have to be defined, e.g.,which method to apply for RNA-analysis,which materials to used, etc; (3) intensive data analysis should be done before using methods of multivariate analysis. The criteria for the presentation of survival curves are given in detail; (4) in multivariate analysis, the Cox proportional hazard regression for survival outcomes is discussed with explanations and examples of the terms relative risk, odds-ratio, hazard and relative hazard. The arrangement and interpretation of a good prognostic study should be performed as an interdisciplinary approach.