Prognostic index models in stage I and II endometrial carcinoma

Background: In most studies, authors recommend the use of independent varia bles in clinical decision-making, but no guidelines are given about how to use all the extracted information. Materials and Methods: By combining two or three prognostically independent variables and using their relative prog nostic impact (results from Cole analyses), a novel way of identifying high - and low-risk groups was developed. A total of 336 women with stage I-II e ndometrial carcinoma of medium or high risk were included. Twenty-one clini co- pathological variables, were initially studied in univariate analyses a nd significant variables were used for the construction of prognostic indic es. Six prognostic indices were constructed, which then were used to calcul ate individual index values for each patient Results: The index values were highly prognostic and used for the identification of a limited high-risk g roup (10% of the patients) and a large low-risk group (90% of the patients) . Indices 1-3 were constructed for pre-treatment situations and included ag e, degree of differentiation nuclear grade and S-phase fraction in various combinations. They identified a high- and a low-risk group with a 5-year di sease-specific survival of 31-37% and 74-79%, respectively. Indices 4 and 5 were constructed using variables available after treatment and index 6 for patients with no evidence of disease (NED) after treatment completion. The 5-year survival for the latter indices were 30-52% and 75-85%, respectivel y. Conclusion: The use of prognostic indices in different clinical situatio ns was generally superior to the use of single variables for the identifica tion of well-defined high- and low- risk groups.