ABSTRACT
BACKGROUND: Clinical and pathological parameters of patients with epithelial ovarian cancer (EOC) do not thoroughly predict patients' outcome. Despite the good outcome of stage I EOC compared with that of stages III and IV, the risk assessment and treatments are almost the same. However, only 20% of stage I EOC cases relapse and die, meaning that only a proportion of patients need intensive treatment and closer follow-up. Thus, the identification of cell mechanisms that could improve outcome prediction and rationalize therapeutic options is an urgent need in the clinical practice. PATIENTS AND METHODS: We have gathered together 203 patients with stage I EOC diagnosis, from whom snap-frozen tumor biopsies were available at the time of primary surgery before any treatment. Patients, with a median follow-up of 7 years, were stratified into a training set and a validation set. RESULTS AND CONCLUSIONS: Integrated analysis of miRNA and gene expression profiles allowed to identify a prognostic cell pathway, composed of 16 miRNAs and 10 genes, wiring the cell cycle, 'Activins/Inhibins' and 'Hedgehog' signaling pathways. Once validated by an independent technique, all the elements of the circuit resulted associated with overall survival (OS) and progression-free survival (PFS), in both univariate and multivariate models. For each patient, the circuit expressions have been translated into an activation state index (integrated signature classifier, ISC), used to stratify patients into classes of risk. This prediction reaches the 89.7% of sensitivity and 96.6% of specificity for the detection of PFS events. The prognostic value was then confirmed in the external independent validation set in which the PFS events are predicted with 75% sensitivity and 94.7% specificity. Moreover, the ISC shows higher classification performance than conventional clinical classifiers. Thus, the identified circuit enhances the understanding of the molecular mechanisms lagging behind stage I EOC and the ISC improves our capabilities to assess, at the time of diagnosis, the patient risk of relapse.
