Fig. 5

Comparison of survival predictive power of the clinical variables and the combined clinical and molecular data. We used the clinical variables (FIGO stage, lymph node metastasis, lymphovascular invasion, stromal depth of invasion, parametrial involvement, and resection margin) for the analysis. During 100 random splits, 80 % of all samples were used to train the model and the remaining 20 % were used as the test set for the C-index calculations. The white box highlights the model built from the clinical variables, and the grey-colored box highlights the models integrating the ANXA2 and ANXA4 data and clinical variables. The combined clinical and ANXA2/ANXA4 model predicting recurrence showed similar predictive power with the clinical variable model (a). However, the combined clinical/molecular-variable model showed better performance than that based only on clinical variables in the model predicting death (Mann-Whitney U-test, p = 0.006) (b). The dashed lines mark the C-index equivalent to a random guess (C-index = 0.5)