Introduction/Background Endometrial cancer (EC) is mostly diagnosed at an early stage with a favorable overall survival. 5-year survival decreases from 95% for localized disease to less than 20% for metastatic disease. Survival is related to known prognostic factors. Nevertheless, they are not sufficient to predict either outcome or recurrence rate/site: to decipher the underlying aberrant biomolecular pathways seems to be promising, although it is not yet applicable in a clinical setting. In that purpose, to investigate EC recurrence patterns according to ESMO-ESGO-ESTRO risk classes, could be beneficial for an early recurrences detection and treatment with a survival rate improvement.

Methodology 758 women diagnosed with EC, and a 5-years follow-up, were enrolled: they were divided into the ESMO-ESGO-ESTRO risk classes (low LR, intermediate IR, intermediate-high I-HR, and high risk HR) and surgically treated as recommended, followed by adjuvant therapy, when appropriated.

Results Recurrences were detected in 19,5%. Higher recurrence rate (RR) was significantly detected (p<0,001) in the HR group (40,3%) compared to LR (9,6%), IR (16,7%) and I-HR (17,1%). Recurrences were detected more frequently at distant sites (64%) compared to pelvic (25,3%) and lymph nodes (10,7%) recurrences (p<0,0001): this trend was evident in all risk classes except for the LR group, where no differences were detected between local and distant recurrences. 5-year distant-free (LR 99%, IR 94%, I-HR 86%, HR 88%) and local-free survivals (LR 99%, IR 100%, I-HR 98%, HR 95%) significantly differ between groups (p<0,0001 and p=0,003, respectively), even when we stratified according to adjuvant therapy (AT) approach. AT did not modify RRs in all risk classes, except for LR group (p=0,01) (85,71% of recurrences occur if no AT was administered).

Conclusion These results strengthen the need to identify biological factors to stratify patients at higher risk of relapse, independently or in addition to their risk class prognosis and current surgical and clinical managements.

Disclosure Nothing to disclose.